Morphosyntactic Expression In Functional Grammar Casper De Groot

Morphosyntactic Expression In Functional Grammar
Casper De Groot download
https://ebookbell.com/product/morphosyntactic-expression-in-
functional-grammar-casper-de-groot-7002256
Explore and download more ebooks at ebookbell.com

Here are some recommended products that we believe you will be
interested in. You can click the link to download.
Morphosyntactic Expression In Functional Grammar Casper De Groot
Editor Kees Hengeveld Editor
https://ebookbell.com/product/morphosyntactic-expression-in-
functional-grammar-casper-de-groot-editor-kees-hengeveld-
editor-50955220
Morphosyntactic Categories And The Expression Of Possession Kersti
Borjars David Denison Alan Scott
https://ebookbell.com/product/morphosyntactic-categories-and-the-
expression-of-possession-kersti-borjars-david-denison-alan-
scott-4731232
Morphosyntactic Variation In Medieval Celtic Languages Corpusbased
Approaches Elliott Lash Editor Fangzhe Qiu Editor David Stifter Editor
National University Ireland Maynooth Editor
https://ebookbell.com/product/morphosyntactic-variation-in-medieval-
celtic-languages-corpusbased-approaches-elliott-lash-editor-fangzhe-
qiu-editor-david-stifter-editor-national-university-ireland-maynooth-
editor-50266840
Morphosyntactic Persistence In Spoken English A Corpus Study At The
Intersection Of Variationist Sociolinguistics Psycholinguistics And
Discourse Analysis Benedikt Szmrecsanyi
https://ebookbell.com/product/morphosyntactic-persistence-in-spoken-
english-a-corpus-study-at-the-intersection-of-variationist-
sociolinguistics-psycholinguistics-and-discourse-analysis-benedikt-
szmrecsanyi-50266890

Morphosyntactic Issues In Second Language Acquisition Danuta
Gabrybarker Editor
https://ebookbell.com/product/morphosyntactic-issues-in-second-
language-acquisition-danuta-gabrybarker-editor-51813576
Morphosyntactic Change A Comparative Study Of Particles And Prefixes
1st Edition Bettelou Los
https://ebookbell.com/product/morphosyntactic-change-a-comparative-
study-of-particles-and-prefixes-1st-edition-bettelou-los-5745606
Morphosyntactic Issues In Second Language Acquisition Danuta
Gabrysbarker
https://ebookbell.com/product/morphosyntactic-issues-in-second-
language-acquisition-danuta-gabrysbarker-1664734
Morphosyntactic Persistence In Spoken English A Corpus Study At The
Intersection Of Variationist Sociolinguistics Psycholinguistics And
Discourse Analysis Benedikt Szmrecsanyi
https://ebookbell.com/product/morphosyntactic-persistence-in-spoken-
english-a-corpus-study-at-the-intersection-of-variationist-
sociolinguistics-psycholinguistics-and-discourse-analysis-benedikt-
szmrecsanyi-1852804
Morphosyntactic Change Functional And Formal Perspectives Olga Fischer
https://ebookbell.com/product/morphosyntactic-change-functional-and-
formal-perspectives-olga-fischer-1936144

Morphosyntactic Expression in Functional Grammar
W
DE
G

Functional Grammar Series 27
Editors
Casper de Groot
J. Lachlan Mackenzie
Mouton de Gruyter
Berlin · New York

Morphosyntactic Expression
in Functional Grammar
edited by
Casper de Groot
Kees Hengeveld
Mouton de Gruyter
Berlin · New York

Mouton de Gruyter (formerly Mouton, The Hague)
is a Division of Walter de Gruyter GmbH & Co. KG, Berlin.
® Printed on acid-free paper which falls within the guidelines
of the ANSI to ensure permanence and durability.
Library of Congress Cataloging-in-Publication Data
Morphosyntactic expression in functional grammar / edited by Casper
de Groot, Kees Hengeveld.
p. cm. — (Functional grammar series ; 27)
Includes bibliographical references and index.
ISBN 3-11-018365-X (hardcover : alk. paper)
1. Grammar, Comparative and general — Morphosyntax. 2. Func-
tionalism (Linguistics). I. Groot, C. de (Casper) II. Hengeveld,
Kees, 1957- III. Series.
P290.M67 2005
415'.9—dc22
2005001606
Bibliographic information published by Die Deutsche Bibliothek
Die Deutsche Bibliothek lists this publication in the Deutsche Nationalbibliografie;
detailed bibliographic data is available in the Internet at <http://dnb.ddb.de>.
ISBN 3-11-018365-X
© Copyright 2005 by Walter de Gruyter GmbH & Co. KG, D-10785 Berlin.
All rights reserved, including those of translation into foreign languages. No part of this book
may be reproduced or transmitted in any form or by any means, electronic or mechanical,
including photocopy, recording or any information storage and retrieval system, without
permission in writing from the publisher.
Printed in Germany.

Preface
Over the last decade, morphological and syntactic issues have received
relatively little attention in Functional Grammar. This is partly due to the
fact that this grammatical model, given its functional orientation, assigns
pride of place to pragmatics and semantics. A further reason is that given
the developments within the theory during this period increasing attention
had to be paid to discourse structure on the one hand and the lexicon on the
other. Yet any fully-fledged theory of language needs a well-defined mor-
phosyntactic component, and there is a revived awareness in the FG-
community of the urgency of the task of developing this component of the
theory in more detail.
Since the original formulation of Functional Grammar in Dik (1978), a
distinction has generally been made between three sets of expression rules:
one set responsible for the form of linguistic units, one for the order of
linguistic units, and one for the prosodic features of linguistic units. The
expression rule component, which contains these three sets of expression
rules, takes abstract pragmatic and semantic underlying clause structures as
its input, and converts these into actual linguistic expressions. Much of the
work on expression rules in Functional Grammar has therefore been dedi-
cated to finding out the exact relationship between underlying pragmatic
and semantic categories on the one hand, and the syntactic, morphological,
and phonological manifestations of these categories on the other, in a wide
variety of languages.
Several of the basic assumptions concerning the organization of the ex-
pression rule component in Functional Grammar have recently been chal-
lenged. First of all, in a number of publications, Bakker (1999, 2001) em-
phasizes the problems posed to the approach summarized above by the
complex interactions that one finds in many languages between the sets of
expression rules that have to account for form on the one hand and those
that establish order on the other. He proposes to deal with these interac-
tions in a radically different integrated implementation of the expression
rule component in FG, called the Dynamic Expression Model. Secondly, in
Functional Discourse Grammar (Hengeveld 2003a, 2003b; Mackenzie and
Gomez Gonzalez eds. 2003; Hengeveld and Mackenzie fc.) a further step is
taken by considering morphosyntactic and phonological representations to

vi Preface
be part of the underlying structure of the grammar rather than as the output
of that grammar.
In view of the issues mentioned in the preceding paragraphs, this book
contains three groups of papers.
The first two articles further develop the idea of the Dynamic Expres-
sion Model in general terms. Dik Bakker's paper concentrates on agree-
ment phenomena, John Connolly's on word order.
The next five articles look at morphosyntactic expression in Functional
Discourse Grammar. Kees Hengeveld shows how the idea of a dynamic
implementation may be applied within Functional Discourse Grammar.
Niels Smit proposes a way to handle noun incorporation within this model.
Casper de Groot stresses the importance of a separate morphosyntactic
level of analysis, and Francis Cornish gives a detailed analysis of the phe-
nomenon of locative inversion. In their analysis of agreement and cross-
reference Anna Siewierska and Dik Bakker explore the possibility of using
Functional Discourse Grammar as a model of the speaker.
The remaining articles provide detailed analyses of a range of semantic
and pragmatic categories and their morphosyntactic expression in a wide
variety of languages. Two papers concentrate on non-verbal predication:
Eva H. van Lier shows that typological hierarchies uncovered in cross-
linguistic research using Functional Grammar can be applied equally well
to developmental language data. Angel Herrero-Blanco and Ventura Sala-
zar-Garcia investigate the appropriateness of the copula support rule in
Functional Grammar for the analysis of Spanish Sign Language. The next
two papers deal with operators of various types: Annerieke Boland pro-
vides a new analysis of operators of tense, aspect, and event quantification
in English, while Ahmed Moutaouakil studies exclamative sentences in
Arabic. The next paper is concerned with the analysis of term phrases:
Evelien Keizer gives a new classification of close appositions in English
using the Functional Grammar framework. The issue of syntactic, seman-
tic, and pragmatic functions in Functional Grammar are the topic of the last
four papers. Arok Wolvengrey discusses inversion in Plains Cree, and Ole
Nedergaard Thomsen obviation in Mapudungun. Johan Lotterman and
Lachlan Mackenzie study the reasons behind the unexpected occurrence or
non-occurrence of an absolutive marker in Tanggu, and Lisbeth Falster
Jakobsen the form of pronominal expressions in Danish.
In all, the contributions to this volume show that the issue of morpho-
syntactic expression in Functional Grammar is very much alive and moving
into promising new directions, while at the same time contributing to a

Preface vii
better understanding of a large number of morphosyntactic phenomena in a
wide variety of languages.
We are grateful to a large number of anonymous colleagues for their
willingness to act as a referee in selecting the papers for this volume. We
are also very much indebted to Lachlan Mackenzie, who went through the
entire manuscript and provided many detailed and helpful comments. Fi-
nally, we would like to thank Niels Smit, who very efficiently and in an
extremely short time span prepared the final version of this book.
March 2005 Casper de Groot
Kees Hengeveld

Contents
Agreement: More arguments for the dynamic expression
model
Dik Bakker 1
Constituent ordering in the expression component
of Functional Grammar
John H. Connolly 41
Dynamic expression in Functional Discourse Grammar
Kees Hengeveld 53
Noun incorporation in Functional Discourse Grammar
Niels Smit 87
Morphosyntactic templates
Casper de Groot 135
A crosslinguistic study of 'locative inversion':
Evidence for the Functional Discourse Grammar model
Francis Cornish 163
The agreement cross-reference continuum:
Person marking in FG
Anna Siewierska and Dik Bakker 203
The explanatory power of typological hierarchies:
Developmental perspectives on non-verbal predication
Eva H. van Lier 249
Non-verbal predicability and copula support rule in
Spanish Sign Language
Angel Herrero-Blanco and Ventura Salazar-Garcia 281

χ Contents
A new view on the semantics and pragmatics of
operators of aspect, tense and quantification
Annerieke Boland 317
Exclamation: Sentence type, illocution or modality?
Ahmed Moutaouakil 351
Close appositions
Evelien Keizer 381
Inversion and the absence of grammatical relations
in Plains Cree
Arok Wolvengrey 419
Direction diathesis and obviation in Functional Grammar:
The case of the inverse in Mapudungun, an indigenous language
of south central Chile
Ole Nedergaard Thomsen 447
Unexpected insertion or omission of an absolutive marker
as an icon of a surprising turn of events in discourse
Johan Lotterman and J. Lachlan Mackenzie 483
Pronominal expression rule ordering in Danish and the question
of a discourse grammar
Lisbeth Falster Jakobsen 503
Index of names
Index of languages
Index of subjects
525
529
531

Agreement: More arguments for the dynamic
expression model
Dik Bakker
1. Introduction
In Bakker (1999; 2001) it is argued that the way in which the Functional
Grammar expression rules (ER) are traditionally organized leads to two
kinds of fundamental problems: undergeneration and overgeneration. There
are a number of arguments for this position. In this first section I will
briefly discuss what I think are the more important ones. For more argu-
ments I refer to the contributions mentioned above.
Undergeneration implies that certain forms that actually occur in lan-
guages cannot be produced systematically by the expression rules. Thus,
the strict organization of ER into two stages, i.e. first the generation of
forms and then their linearization, makes all constellations in which form is
in some way dependent on order impossible sui generis. The interdepen-
dency between form and order is in fact quite common in the languages of
the world, above all in the case of agreement phenomena in languages with
a non-rigid constituent order.1
1 will give four examples of this. In (1), from
Basque, in prenominal relative clauses the auxiliary verb agrees with the
antecedent in person and number (la); however in postnominal ones it
agrees in defmiteness, number and case (lb). In (2), from Selknam, a lan-
guage spoken on Tierra del Fuego, now probably extinct, the order of the
Subject and Object agreement markers on the verb is the mirror image of
the order of the corresponding constituents vis ä vis the verb while both
orders are allowed. In (3), from Koegu, a Surmic language from Ethiopia,
the verb agrees optionally with the subject when it is postverbal (3a/b), but
obligatorily when it is preverbal (3c/d). In (4), from Yagua, a language
from the Amazon area, there is an agreement marker on the verb in the case
of a postverbal subject (4a) but not in the case of a preverbal subject (4b).

2 Dik Bakker
(1) Basque (Saltarelli 1988:36)
a. Gizon-ari haragi-a sal-du
Man-SG.DAT meat-SG.ABS sell-PRF
d-io-0-n harakin-ari
3.ABS-3.SG.DAT-3.SG.ERG-COMP butcher-SG.DAT
hortz-ak eror-i za-izk-io.
tooth-PL.ABS fall-PRF AUX-ABS.PL-3.SG.DAT
'The teeth have fallen out on the butcher that has sold the meat to
the man.'
b. Galtzerdi hori-ek soin-ean
Stocking that-PL.ABS body-SG.LOC
d-a-u-z-ka-zu-n-ak
3.ABS-PRS-have-PL.ABS-have-2.SG.ERG-COMP-PL.ABS
polit-ak d-i-ra.
beautiful-PL.ABS 3.ABS-PRS-be
'The stockings which you have on, are very pretty.'
(2) Selknam (Nachlis 1973:17)
a. P'ejj kah-jqe-en mer na?
Knife M-see-F PST woman
b. Na? tah-jqe-enn mer p'ejj.
Woman 3.SG-see-M PST knife
'The woman saw the knife.'
(3) Koegu (Hieda 1998:367)
a. A-mat-i-yaa aan.
l.SG-drink-3.SG l.SG
b. 0-mat-aa aan.
drink-3.SG l.SG
c. Aan a-at-i-yaa.
l.SG l.SG-drink-3.SG
d. *Aan 0-at-i-yaa.
l.SG drink-3.SG
Ί drank it.'

More arguments for the dynamic expression model 3
(4) Yagua (Payne 1990:30)
a. Sa-juuy Anita.
3.SG-fall Anita
b. Anita 0-juuy,
Anita 3.SG-fall
'Anita fell.'
Though there may be solutions for such phenomena based on information
contained in underlying representations only, i.e. purely on semantic or
pragmatic factors, examples from Arabic to be discussed later show that
not all problems of form may be solved before the linear order rules are
executed.2
A second source of undergeneration is the fact that there are no clear
principles for the expansion of operators apart from centripetal expansion,
i.e. operator application according to the inside-out order in which they are
are found in underlying representations (cf. Dik 1997:381f). If this iconic-
ity-driven principle of expression were to be the only device for the genera-
tion of grammatical material, we would have a problem in dealing with
underlying representations like the one in (5).
(5) (int Ei : (X] : (past ej : (smile [V] (dl Xj : John [PN])Ag))))
If we expanded the operators in (5) in a straightforward centripetal fashion,
we would first derive a past form for the verbal predicate before we gener-
ated the auxiliary do on the basis of the Interrogative operator at the illocu-
tionary level, which would lead to the ill-formed utterance in (6a) instead
of the correct one in (6b).
(6) a. *Do John smiled?
b. Did John smile?
So, in this case the illocutionary operator should trigger before the tense
operator does, in contrast to their centripetal order in the underlying repre-
sentation. Obviously, there must be more principles at work here to decide
which combinations of what operators might be selected at what stage of
the form generation process, be they universal, typological or language
specific.3
To my knowledge, no suggestions for other such principles are to
be found in the FG literature to date, nor do we find any complete example

4 Dik Bakker
of the derivation of a sentence based just on centripetality. The fact that
there is no worked-out strategy for combining expression templates at the
respective syntactic levels probably plays an important role in this respect.
In other words: there is no theory of syntactic constituency. As such, un-
dergeneration threatens the most elementary of adequacy requirements of a
linguistic theory, i.e. observational adequacy.
Turning to overgeneration, this is essentially due to the lack of con-
straining principles and formal structures to apply them to. Since there
seems to be no way of controlling which (combination of) operators oper-
ate on which operands, which outputs may and may not be recycled to later
rule applications, etc., the impression at least is that 'anything goes'.4
Without enough constraining principles on the semantics-syntax interface
and given the ban on filtering devices in FG we will in all likelihood be left
with a large number of potentially meaningful but formally unacceptable
utterances apart from the well-formed ones. This in turn threatens the cog-
nitive and typological adequacy of the theory; it also casts doubt on the
learnability of a grammar operating with the traditional expression rules.
In order to solve these problems, at least in principle, Bakker (2001)
formulates a version of the expression component which integrates the
form generation and linearization modules, thus providing a solution for
cases such as those illustrated above. Furthermore, in this newly shaped ER
component, the rules are assumed to operate in a dynamic fashion. They
produce tree-like structures with phonological strings for their terminal
nodes. As in the case of the original ER component the point of departure
is that the expression rules are triggered by the interaction between the
semantics and pragmatics of the underlying representation. The well-
known templates and placement rules reappear, albeit in a slightly modi-
fied form. According to Bakker & Siewierska (2002) the dynamicity of
rule execution is governed by five principles of tree construction. Three
principles determine the way expression trees come into existence, viz. via
top-down, left-to-right, and depth-first expansion. Two more principles
determine the way in which linguistic information is passed on during the
dynamic expansion process, viz. inheritance and percolation. The overall
view will be that the shape of the expression rules in a synchronic grammar
of a language L is dictated not only by the requirements of underlying rep-
resentations, but also by the direct and indirect influence of non-linguistic
factors of a cognitive nature, such as iconicity and economy. Is is also as-
sumed that the dynamic process itself plays an active role in instances of

grammaticalization, as an active factor or, diachronically, as an independ-
ent ground for explanation.5
As has been argued in the contributions referred to above a large class
of problems of the type demonstrated by examples (1) - (5) may well be
solved within this dynamic ER framework. In section 3 below this will be
demonstrated for a domain where form and order interdependencies are
rather common in the languages of the world, i.e. the shape of subject
agreement marking on the verb. However, certain problematic cases from
this domain remain unsolved if only the three tree expansion principles top
down, left to right, and depth first are available. On the basis of a case
study of subject agreement in Arabic it will be argued, in section 4, that a
fourth tree expansion principle is called for, i.e. a limited amount of look
ahead. Before we launch into this in section 2, there will be a short resume
of the dynamic expression rule component as it has been developed to date.
However, a remark on linguistic modelling is called for first. According
to Dik (1997:1), the ultimate question FG wants to answer is how the lan-
guage user works. From this it follows that the requirement of cognitive
adequacy which the theory imposes on itself should be taken very seri-
ously.6
Interestingly, however, the models that have been presented so far
within FG theory model the grammar rather than the speaker and/or the
addressee (cf. Dik 1997:60; Hengeveld 2004). I would argue that such
models may serve a purely organizational purpose, such as structuring the
theory, ordering the respective components of the grammar in relation to
each other, clarifying a linguistic discussion or even planning the linguistic
enterprise itself. But grammar models, which were not constructed directly
on the basis of insights from the psychology of language cannot by impli-
cation serve to test the theory on its cognitive adequacy. I would therefore
propose the development of a model of the language user, arguably even
distinguishing between a model of the spreaker and one of the addressee.7
Elsewhere (Siewierska & Bakker, this volume) it is shown that the dy-
namic expression rule model fits well into a dynamic discourse-oriented
model of the speaker based on FG theory. Having said this, for the current
discussion we will assume, as in the traditional transformational grammar
models, including the standard FG grammar model, that a fully specified
underlying representation, with all the semantic and pragmatic aspects
filled in, is available before the first expression rule is triggered. Discourse
representations will play no significant role here.

6 Dik Bakker
2. The dynamic ER component: A short introduction
The idea of dynamic expression hinges on the integration of the form gen-
eration rules and the templates and placement rules of the standard theory
into a tree-like structure that represents the history of the expression of a
specific underlying representation (URj). Such a tree should be seen as the
result of a dynamic process rather than as a static representation of the
(morpho-)syntactic structure of the corresponding URj as in most syntactic
theories. The tree consists of a number of nodes, which should be inter-
preted as the constituents of the tree. Each node contains a specific part of
the URj under expression including its lexical material, operators, functions
and layered structure. Examples are the subject term, the main predicate,
an adjectival restrictor or a relative clause. A node explicitly specifies the
primary and auxiliary μ operators relevant at the corresponding stage of the
expression process. In fact, a node may only contain μ operators and no
lexical material at all. On the basis of its constellation of lexical and
grammatical material, a node selects a template from the grammar, which
specifies the linear order of the actual elements of the URj element handled
by the node. For each of the slots in that template a new daughter node is
created, in a left-to-right fashion, which then will deal with the element of
the mother node that has been assigned to it, and so on recursively. This
process continues until each lexical element resulting from the original URj
has been assigned its own terminal node. Unlike in the case of the standard
model of expression, grammatical elements such as articles, auxiliaries and
plural suffixes are not created before but rather during the process of lin-
earization, at the moment at which precisely enough information is avail-
able in order to select the right form and in time for them to introduce fea-
tures that derive from them, such as case. Typically, grammatical markers
are the expression of a terminal node which contains only μ operators and
no lexical material whatsoever. The corresponding grammatical element is
then selected by the grammar. Some terminal nodes may even be prespeci-
fied for a form, which is then considered to be completely grammatical-
ized. This may be the case for bound forms such as plural suffixes, which
only have phonologically determined allomorphs.
This top-down process starts with the creation of a node for the fully
specified underlying representation under expression, typically a clause,
but it may also be a term or any other well-formed underlying construct.
For the English sentence in (7) below the top node for its expression may
look as in (8). In node specifications, Slot is the lable that identifies the

node in the grammar. It corresponds to the notion of functional slot in the
traditional order templates. Cat contains the morpho-syntactic category of
the constituent to be expressed via this slot. Config is the field that pre-
specifies the type of configuration to be expressed by this type of node (in
this case a clause). The set of all possible configurations that fit this de-
scription could be seen as the (recursive) defmition-by-enumeration of the
Slot lable. FncFtrs and FrmFtrs are the functional (i.e. primary) and formal
(i.e. auxiliary) μ operators relevant for this node. SubCat determines the
subcategories of Cat, and will contain the template that determines the
order in which the subconstituents will be expressed. It could be seen as
the (recursive) definition-by-enumeration of the Cat lable, and therefore as
the definition of morphosyntactic constituency in the corresponding lan-
guage.8
(7) [decl E,:[Xi:[pres e^'smart'[A] (rempi Xi:'girl'[N])zer0,subject ] ] ] ]
'Those girls are smart.'
(8) Node k (uninstantiated)
Slot: clause
Cat: s
Config: [ILLOC [ [TENSE [ PRED CAT ARGi ]]]]
FncFtrs: ILLOC, TENSE, CAT, PERSON.subject,
NUMBER,subject
FrmFtrs:
SubCat:
(8) gives the node in its abstract, uninstantiated form, the way in which it is
part of the language-specific grammar. It is selected on the basis of certain
formal aspects of the UR under expression, in this case a clause. The Con-
figuration field of this node will now be filled by the UR of (7), which
leads to the partially instantiated version of it in (9) below. Note that the
two types of feature sets are instantiated at the same time, to the extent that
their values are available.

8 DikBakker
(9) Node 1 (partially instantiated)
Slot: clause
Cat: s
Config: [decl Ei:[Xi:[pres ei:['smart'[A] (rem pi x^'girl'
[N])zero,subject ]]]]
FncFtrs: decl, pres, [A], [-sp,-hr], pi
FrmFtrs:
SubCat:
In the following step, the right template will be selected for the expression
in the Config field of Node 1. This will give us its fully instantiated version
in (10).
(10) Node 1 (fully instantiated)
Slot: clause
Cat: s
Config: [decl Ei:[Xi:[pres ei:['smart'[A] (rempi x^'girl'
[N])zero,subject ]]]]
FncFtrs: decl, pres, [A], [-sp,-hr], pi
FrmFtrs:
SubCat: PI, subject, vfin, mainpred
The process now continues in a left-to-right, top-down fashion with the
expansion of the PI position. For this slot several alternative nodes will be
available, a situation comparable to the alternative placement rules for PI
in the standard theory (cf. Dik 1997:421). The right candidate node will
then be selected, and the corresponding element of the Config field of node
1 will be inserted into the Config field of this newly created node 2 - in
this case the subject term. The process will continue depth-first by select-
ing the right template to express the subject term, here a noun phrase tem-
plate. Subsequently the leftmost element of this template - in this case a
determiner position - will be expanded. The final result of this process will
be a tree as in figure 1 below. The figures next to the branches indicate the
order in which node expansion takes place. Note that the tree is not just a
representation of the syntactic distribution of the material in the UR but
extends into the morphological level.

'girl' 's' 'smart'
Figure 1. Full tree expansion
The typical way in which the functional and formal features of nodes get
their values is by inheritance. These values are either derived directly from
the UR material in the configuration field or copied from the Config field
of the mother node. However, feature values may also be acquired via
processes that take place in daughter nodes of the node under discussion. In
those cases, features will percolate bottom upwards. The percolation
mechanism is necessary for those cases in which formal information which
results from early and very 'deep' stages of expression, such as case as-
signment by an adposition, certain order phenomena or the phonological
form of some grammatical marker, will remain available for later stages in
the expression process, where it may have some impact on the form.9
Trees such as those in figure 1 may also be seen as static constituent
trees, as in Role and Reference Grammar (Van Valin & La Polla 1997) or
Government and Binding theory (Haegeman 1994). However, unlike the
former, dynamic ER trees are not stored as ready-mades in the fund, and
selected on the basis of underlying semantic representations. And unlike

10 DikBakker
the latter, they are not the result of a transformational process that leads
from a base-generated structure to a surface structure and which is in fact
unrelated to the actual process of uttering a sentence. FG expression trees
are the dynamic result - one could say: the side-effect - of the actual ex-
pression process, and exist only after an utterance has been completed. In a
way, they are the utterance itself.
This short introduction should be enough for our purposes here. For a
more comprehensive treatment we refer to the articles mentioned. In the
sections below, several finer details will be discussed, as well as a neces-
sary extension of the 5 principles mentioned above.
3. Attempting to solve the form-order problem
Form-order problems for which a unified underlying factor may be found
can be handled in a more or less straightforward fashion. This situation
arguably applies to the following case from Konjo, an Austronesian lan-
guage from Sulawesi.
(11) Konjo (Friberg 1996:141, 146)
a. Na-peppe'-i Amir asung-ku.
3.ERG-hit-3.ABS Amir dog-POSS.l
'Amir hit my dog.'
b. I-nai ang-kanre-i lamejaha-ku?
PI-who TRANS.DEF-ate-3.ABS sweet.potatoes-POSS.l
'Who ate my sweet potatoes?'
c. I-Ali ang-kanre-i lamejaha-ta
PI-Ali TRANS.DEF-ate-3.ABS sweet.potatoes-POSS.2
'Ali ate your sweet potatoes.'
In (1 la) we have the pragmatically neutral declarative VSO order in which
the verb displays agreement with both the Agent and the Patient. In (1 lb) -
a question - and (11c), the answer to that question, the focalized constitu-
ent is placed in preverbal position. At the same time the agreement marker
for the Agent is replaced by a marker which codes transitivity and the defi-
niteness of the preverbal focus constituent. So, we could reason that in this
case the word order and the type and form of the Agent agreement prefix

are determined by the same underlying factor, and therefore in some sense
are the expression of it, i.e. of the pragmatic status (the information focus)
of the Agent, among other features.
A somewhat more complex form-order relationship is illustrated in the
following example from Fore, a language from Papua New Guinea.
(12) Fore (Foley 1986:171)
a. Yaga: wä a-egü-i-e.
pig man 3SG.PAT-hit-3SG.AG-DECL
'The man hits the pig.'
b. Yaga:-wama wä a-egü-i-e.
pig-ERG man 3SG.PAT-hit-3SG.AG-DECL
'The pig hits the man.'
c. Wa mäsi a-ka-i-e.
man boy 3SG.PAT-see-3SG.AG-DECL
'The man saw the boy.'
d. Mäsi wä-mä a-ka-i-e.
boy man-ERG 3SG.PAT-see-3SG.AG-DECL
'The man sees the BOY.'
The order of the preverbal constituents in this verb final language is rela-
tively free: both APV and PAV orders are acceptable, and probably reflect
pragmatic differences. Often the verbal morphology indicates which con-
stituent is the Agent - the suffix - and which the Patient - the prefix. If this
is ambiguous, an animacy hierarchy determines what is the more likely
Agent, as in (12a). If the speaker wants to assign the Agent role to the ref-
erent which is lower on the hierarchy, as in (12b), then explicit ergative
marking of the Agent is necessary. When both arguments score equal on
the hierarchy, and there is no ERG marker, constituent order is decisive, as
in (12c), which has only one interpretation. If pragmatic reasons make
APV order undesirable, then again case-marking may be called in to dis-
ambiguate, as in (12d).
Example (12) makes clear that the interplay between form and order
phenomena may sometimes be quite complex, and that both are not neces-
sarily the direct expression of the same underlying factor. This makes cases
like the one from Fore at best very problematic for the standard expression

12 DikBakker
rules, especially when order has an impact on form rather than the other
way around. To find out to what extent the dynamic expression rules can
deal with such matters we will have a look at subject-verb agreement in
Arabic, which presents us with a number of relevant problems. Let us first
consider the relatively simple examples in (13) below.
(13) Arabic (Mohammad 2000:109)10
a. ?al-?awladu dza?-u.
the-boysTOp came-3PL.M
b. Dza?-a l-?awladu.
came-3SG.M the-boysFoc
'The boys came.'
In (13a) the subject is topic and we have SV order. In this case there is full
Person/Number/Gender agreement marking on the verb. In (13b), however,
with focal subject and VS order, there is no Number agreement, only Per-
son and Gender. Potentially, this case could be handled by the standard
rules. We would first calculate the form of the verb, taking into considera-
tion the pragmatic function of the subject term. We might have the - sim-
plified and very partial - expression rules for the form of the finite verb in
Arabic as in (14) below. They take the form suggested in Dik (1997:383),
with the pragmatic function of the subject as a condition.
(14) Arabic; form of finite verb
a. TENSE, PERSON, NUMBER, GENDER [PRED [V]] = FORM
Condition: PragF.subject = topic
b. TENSE, PERSON, NUMBER sg,
GENDER [PRED [V]] = FORM
Condition: PERSON = 3, PragF.subject=focus
The arrow in rule (b) is a shorthand for a redundancy rule which inserts
'sg' for the value of Number in the set of μ operators. Subsequently,
placement rules which also take the pragmatic function of the subject into
consideration would then locate the subject term either in the PI template
slot (PragF.subject=topic) or in a postverbal S or Focus slot
(PragF.subject=focus).

More argumentsfor the dynamic expression model 13
Although this looks quite straightforward, there is a fundamental prob-
lem with this approach. It includes the pragmatic function of the subject in
the formulae that calculate the form of the verb. If we do not interpret ex-
pression rules as being mere description, but rather in terms of the underly-
ing elements they really express - provided that they are not completely
arbitrary -, then having a rule like (14) would imply that in the grammar of
Arabic the form of the subject agreement marker on the verb actually ex-
presses the fact that the subject is topic or focus, and is not just an indirect
side effect. This does not seem to be a very satisfactory interpretation of
this constellation, the more so since it works only for third person plural
subjects and not for first and second person, as shown in (15) below.11
In
(15a) we have a preverbal first person plural pronoun. In (15b) it is post-
verbal, a rather marked but not impossible constellation, according to Olm-
sted Gary & Gamal-Eldin (1982:54). In both cases, the marker on the verb
is the one for 1st
plural. The constellation in (15c), with 1st
singular on the
verb is not acceptable.
(15) a. niHin yii-na.
1PL come.PST-IPL
yii-na niHin.
come.PST-IPL 1PL
*yii-t niHin.
come.PST-lSG 1PL
'We came.'
Let us now see how the examples of (13) would be treated in the dynamic
expression rule framework introduced in section 1. In the representations
given below no attempt is made at completeness. They just serve to clarify
how the dynamic expression rules might handle such cases. Any feature
that does not seem to be relevant for the current discussion will therefore
be left out.
The uninstantiated node in (16) might be a candidate node for the sen-
tence level.

14 DikBakker
Slot: clause
Cat: s
Config: [ILLOC [ [TENSE E-VAR [ PRED CAT ARGi ]]]]
FncFtrs: TENSE, PERSON, NUMBER, GENDER
Instantiation for the sentence with subject-topic in (13a) gives the version
in (17).
Slot: clause
Cat: s
Config: [decl [ [past ej : [dza? [V] [def pi X] : ?awladu
[N]]agent .subject,topic ] ] ] ]
FncFtrs: past, PERSON, NUMBER, GENDER
Full instantiation of this node will give us (18).
Slot: clause
Cat: s
Config: [decl [ [past ei : [dza? [V] [def pi Xi : ?awladu
[ N ] ]agent,subject,topic ] ] ] ]
FncFtrs: past, PERSON, NUMBER, GENDER
SubCat: PI, vfin, focus
This node is still only partially instantiated. Three of the four functional
features still have no value.12
The next step is the expansion of the PI node. Assuming that there are
several candidate fillers for this node, one of the alternatives for it may
look like (19) below. Note that the configuration is prespecified for the
subject-topic combination.
Slot: PI
Cat: np
Config: [DEF NUM X j : PRED [N] ]SEMF,subject,topic
FncFtrs: DEF, PERSON, NUMBER, GENDER

After instantiation we get (20). It is assumed that there exists a lexical rule
which finds the value for the 'invisible' Gender feature by looking into the
lexical entry of the head noun.'3
(20) Node 2 (instantiated)
Config: [def pi xj: '?awladu' [N] ]agent,subject,toPic
FncFtrs: def, 3, pi, m
SubCat: art, noun
In a few more steps, which we will not explore further here, this will lead
to the expression of the subject term, including the cliticized definite arti-
cle and the plural suffix. What is crucial is that at node number 2 the rele-
vant features of the subject term are uniquely determined, and therefore
become available. Through percolation the corresponding features of Node
1 will be replaced by these values, which gives us a fully instantiated ver-
sion for this node, as in (18')
(18') Node 1 (fully instantiated)
Slot: clause
Cat: s
Config: [decl [ [past ei : [dza? [V] [def pi Xi : ?awladu
FncFtrs: past, 3, pi, m
SubCat: PI, vfin, focus
The process continues with the selection of a node for the functional vfin
slot. This vfin slot will inherit the values for its Person, Number and Gen-
der features from node 1, and then pass them on to the actual nodes that
express the finite verb form. Since after this step all material in the Config
field of node 1 will be expressed the process is complete and comes to a
halt.14
Although this schema seems to work for the SV case, it is not ade-
quate for the VS one. Person and Gender information is necessary to get
the right verb form, and will not be available to the vfin node if the subject
term has not been expressed yet. This implies that the values for these fea-
tures should be available at the clause level, before the vfin node is created.
This calls for a redefinition of the clause node. In (16') a modified version
is to be found which provides them.
Slot:
Cat:
PI
np

16 Dik Bakker
(16') Node k (uninstantiated)
Slot: clause
Cat: s
Config: [ILLOC [ [TENSE [ PRED CAT ARG, ]]]]
FncFtrs: TENSE, PERSON.subject, NUMBER,
GENDER,subject
With the extra instruction that the values for Person and Gender should be
retrieved from the subject term the values for these features will become
available at the clause level as soon as Config is instantiated. For the ex-
pression of (13a) this has no repercussions. The same steps apply as the
ones given in (17)-(20) above. The only difference is that there will be no
percolation of the Person and Gender values; they are simply available
from the start. However, for the expression of (13b) these subtle changes
are crucial. Since in the grammar of Arabic there is no PI node which is
specified for a focal subject, or indeed for any of the material in the under-
lying representation of (13b), expression will start with the instantiation of
the vfm node. This node will inherit the values for its Person and Gender
features from Node 1. The fact that the Number value remains uninstanti-
ated will trigger a default rule which assigns singular to it.15
This will give
us the right form for the finite verb. The default assignment process is lo-
cal, i.e. there is a barrier to the percolation of the values that were assigned.
After expression of the finite verb, for which there may be two more nodes
to be expanded, one for the verb root and one for the agreement suffix,
expression will continue with the focus node. One of the possible alterna-
tives for this node will look like (21) below.
Slot: focus
Cat: np
Config: [DEF NUM X j : PRED [N] ]SEMF,SYNTF,FOCUS
FncFtrs: DEF, PERSON, NUMBER, GENDER
Since there is an unexpressed candidate for this node, it will be instantiated
as in (22) below. On this occasion, the Number feature will get its value
from the inserted term, i.e. plural. The Person and Gender features inherit
their values from Node 1. The latter values will harmonize with the local
values for these features for obvious reasons.16

(22) Node 5 (instantiated)
Slot: focus
Cat: np
Config: [def pi xj: '?awladu' [N] ]agent,subject,focus
FncFtrs: def, 3, pi, m
SubCat: art, noun
For the examples of agreement in (13) this treatment seems to be satisfac-
tory. It is no longer necessary to assume that form and order phenomena
are the expression of the same underlying factor, in this case topicality
versus focality of the subject term. Rather, we can say that the constituent
order is pragmatically determined while the form of the agreement marking
may be seen as a side-effect of the dynamics of expression. If we express
the subject first, as in (13a), its Number feature becomes available for its
own expression. As a result it will percolate upwards, and become avail-
able for the rest of the expression. This is not the case, however, for (13b).
As for the Person and Gender features of the subject, I will assume that
they are so prominent that they are accessible at the level of the clause.
This accessibility of Person and Gender is coded in (16') via the set of
features that may be instantiated on the basis of the local Config field. As a
result, the corresponding values are available under any of the order vari-
ants, as opposed to the value for Number, which is assigned via a local
default rule and does not percolate.17
Interestingly, this analysis seems to get support from a related aspect of
verbal agreement in Arabic. Unlike many other languages, in case of the
presence of an auxiliary verb, both the latter and the main verbal predicate
agree with the subject. However, when the subject is expressed after the
auxiliary but before the main verb there is person agreement on the latter
but not on the former. This is exemplified in (23a) and (23b), respectively.
(23) Arabic (Bahloul & Harbert 1993:16)
a. Al-bint-aani kaan-ataa ta-ktub-aani darsa-humaa.
The-girl-3.DU was-3F.DU 3F-write-DU lesson-F.DU
b. Kaan-at al-bint-aani ta-ktub-aani darsa-humaa.
was-3F.SG the-girls-3.DU 3F-write-DU lesson-F.DU
'The two girls were writing their lesson.'

18 DikBakker
These facts may well be explained on the basis of the assumption that
number only becomes available for reasons of agreement after the subject
has been expressed and the number feature of the term has been brought
into play.18
Though possibly satisfactory for the above cases, the solutions pre-
sented so far will not suffice to handle other aspects of subject verb agree-
ment in Arabic. Further provisions need to be made, to which we now turn.
4. The necessity of looking ahead
In section 3 we saw that the dynamic top-down, left-right, depth-first ex-
pansion of nodes, combined with the inheritance and percolation of fea-
tures was adequate for the representation of the form-order interaction
phenomena in example (13). The dynamicity aspect even provides a possi-
ble explanation for them. However, this is certainly not the full picture of
subject-verb agreement in Arabic. Complications arise in the case of sub-
jects that consist of conjoined noun phrases with different genders, a prob-
lem labelled 'resolution' in Corbett (1991:261), where it is discussed ex-
tensively. Let us first consider the preverbal situation as exemplified in
(24). Unless stated otherwise, the Arabic examples below are from the
discussion in Mohammad (2000:112ff).
(24) a. ?al-?awladu w-al-banatu qara?u kitaban.
the-boys and-the-girls read.3PL.M book
'The boys and the girls read a book.'
b. ?al-banatu w-al-?awladu qara?u kitaban.
the-girls and-the-boys read.3PL.M book
'The girls and the boys read a book.'
As these examples show, gender conflicts are resolved in favour of mascu-
line, independent of the order of the conjuncts. Only when all conjuncts are
feminine do we get feminine agreement on the verb. With VS order, how-
ever, the situation is more complex. In such a case, the order of the con-
juncts is decisive, as shown in (25) below. Note that, unlike the situation
with preverbal subject we have agreement with the conjunct closest to the
verb, masculine for (25a) and feminine for (25b). In the latter case mascu-
line agreement is even ungrammatical, as shown in (25c). Note furthermore

that also in these cases we have the singular rather than the plural form on
the verb. Apparently, the conjunction is not an extra factor for the assign-
ment of plural number in VS sentences.
(25) a. Qara?a ?al-?awladu w-al-banatu kitaban.
read.3SG.M the-boys and-the-girls book
b. Qara?at l-banatu w-al-?awladu kitaban.
read.3SG.F the-girls and-the-boys book
c. *Qara?a l-banatu w-al-?awladu kitaban.
read.3SG.M the-girls and-the-boys book
This situation changes, however, when there is a constituent between the
verb and the subject. In the sentences of (26) we have VXS order, with a
feminine subject. Although (26a) is acceptable, (26b) is observed more
often.
(26) a. qad kan-at yaskun-ut
PRT was-3SG.F settle-3SG.F
l-iraq-a Pumamun muxtalifatun.
the-Iraq-ACC people.F.PL different
b. qad kan-a yaskun-u
PRT was-3SG.M settle-3SG.M
l-iraq-a Pumamun muxtalifatun.
the-Iraq-ACC people.F.PL different
'Different peoples had settled in Iraq.'
Apparently, the distance between the expression of the verb form and the
(first conjunct of the) subject determines the amount of semantic corre-
spondence between the agreement marker and the source of the agreement.
The greater the distance, the less the semantics of the source are accessible.
Other semantic factors may play a role in such processes as well. Cor-
bett (1991) analyses examples of subject-verb agreement in Classical Ara-
bic VS sentences as determined by the (in)defmiteness of the subject term.

20 DikBakker
Below, in (27a) there is feminine agreement followed by a definite femi-
nine subject; in (27b) there is masculine agreement followed by an indef-
nite feminine subject. Note, however, that in the latter case there is also an
intervening (definite) object as in the examples of (26) above.
(27) Classical Arabic (Corbett 1991:125)
a. Jä-at Hindun.
came-F.SG Hind (female proper name)
'Hind came.'
b. Hadar-a l-qädiya ('i)mra'atum.
came.before-M.SG the-judge woman
Ά woman came before the judge.'
The examples in (24)-(27) are highly suggestive of the following scenario
for agreement resolution with conjoined subjects in Arabic. In SV situa-
tions there is no real resolution problem. The agreement is masculine, pro-
vided that at least one of the conjuncts is masculine, otherwise it is femi-
nine. In VS situations, apart from the general unavailability of Number as
an agreement factor for third person subjects, the conjunct to be expressed
first determines the gender of the agreement marker on the verb, unless
there is intervening material. In that case Gender becomes inaccessible as
well and masculine gender serves as the default. Finally, definiteness may
add to the accessibility of a subject term. That definiteness may also play a
role in the form of agreement marking in other languages may be illus-
trated on the basis of the example in (28) from Hua, a Papuan language. In
(28a), with an object marker on the verb, the corresponding referent should
be interpreted as definite; in (28b), without such a marker, we are dealing
with an indefinite interpretation.
(28) Hua (Haiman 1980:371)
a. Vedemo p-go-e.
men 3PL-see-lSG
Ί saw the men.'
b. Mna-vza-mo ko-e (*p-go-e).
bird-COLL-PL see-lSG (3PL-see-lSG)
Ί saw birds.'

Assuming that this analysis of the agreement facts in Arabic is correct and
that it should be dealt with within the expression component, the set of
principles which determine what information is available at what stage
during expression needs to be modified. The combination of depth first and
inheritance which constrained the agreement marking in section 3 does not
suffice to code the facts of this section. What we need is a mechanism that
allows the expression rules to look ahead, i.e. to inspect features of a con-
stituent that has not been expressed yet but is about to be expressed. This
mechanism should represent the accessibility of features in a more subtle
way than the yes/no fashion of node 1 in example (16'), repeated here for
convenience.
(16') Node k (uninstantiated)
Slot: clause
Cat: s
Config: [ILLOC [ [TENSE [ PRED CAT ARGi ]]]]
FncFtrs: TENSE, PERSON.subject, NUMBER,
GENDER,subject
In the case of (16'), the Person and Gender values are available immedi-
ately upon instantiation of the clause node; they are maximally accessible.
The value for Number becomes available only upon the expression of the
element that contains it, here the subject. However, given the facts of Ara-
bic, we have to cater for (at least) three types of accessibility of semantic
and pragmatic information in underlying representations.
- Global accessibility: the information is available immediately after the ex-
pression process has started.
- Relative accessibility: the information is available some time before the
expression of the UR constituent to which it belongs but not for the com-
plete expression of the clause concerned
- Local accessibility: the information is available only upon the expression of
the UR constituent to which it belongs.
In terms of subject-verb agreement in Arabic, Person would be an example
of global accessibility, Gender of relative accessibility and Number of lo-
cal accessibility. Obviously, relative accessibility is a dynamic notion since
it introduces time into the expression schedule. Rather than measuring
relative accessibility in time units, such as milliseconds, however, it seems

22 Dik Bakker
to be more adequate in our model to measure it in terms of a 'look ahead
window'. This window represents the maximum number of nodes to the
right of the node currently under expression that may be taken into consid-
eration in order to find the relevant feature values for the current node.19
An adequate representation of accessibility in expression schemata may be
a scale, with values between 0 (local accessibility) and oo (global accessi-
bility). In order to constrain the accessibility of the respective features in
node representations, the accessibility window may then be added to the
features involved. In an extended version of the formalism, there might be
features that co-determine the width of the accessibility window of some
feature, such as defmiteness in the case of gender in Arabic, as in example
(27) above.20
Taking all this into consideration, we may modify the defini-
tion for the clause node as in (16") below.
(16") Node k (uninstantiated)
Slot: clause
Cat: s
FncFtrs: TENSE, PERSON.subject <od>, GENDER.subject <n>,
NUMBER,subject <0>
The descriptor <od> is an instruction to find a value for this feature in the
Config, whatever its level of embedding.21
The descriptor <0> implies the
opposite. The corresponding feature value should be present locally, i.e. it
should be found at the top level of the Config constituent. For a clause, this
would only concern the illocutionary operators and satellites; for a term it
would at least concern its pragmatic, semantic and syntactic functions, the
ω operators, and some features of the nominal head, such as its Gender and
its nominal aspect in terms of Rijkhoff (2002). Since <0> is the strongest
constraint, it may well be the best candidate for the default in node descrip-
tions. Any descriptor <n>, where 0 < η < oo, implies that the corresponding
feature, if not readily available, should become available after up to a
maximum of η nodes have been created to the right of the current node at
the level of the same SubCat.
The <oo> and <0> descriptors may be handled adequately under the
original 5 principles, but <n> calls for an extension. In order to implement
<n> the model has to allow for a limited amount of look ahead. Thus, to
ensure the correctness of a form, it may be necessary to access information
from a constituent which will be expressed either immediately after the

current node or to a maximum of η nodes away. This is indeed a special
example of the intertwining of form and order. For the model this means
that, at least in these cases, the development should be breadth first, at least
one step ahead, rather than depth first. Arguably, this is the reflection in
the model of parallel processing by speakers. In a really parallel model,
while one processor works on the expression of the constituent in node k,
another process is busy in the background selecting the constituent for
node k+1. Some information from the latter will then be available for the
process under node k before it terminates and node k+1 moves to the fore-
ground. If we translate this parallelism into a limited amount of breadth-
first node expansion, we get a dynamic tree that looks more or less like the
one in figure 2 below. Of course, such a 'flat' representation can never
reveal the real intricacies of parallel processing.22
In such a setup it must be
assumed that, after the subject node 3 has been created and partially instan-
tiated, the top-down process for node 2 has progressed, however not to the
extent that the verb form has been expressed fully. Matters might be even
more complex in the sense that there may be more than two parallel proc-
esses.
Figure 2. Partial tree with breadth-first development
Two points need some extra attention here. Firstly, on several occasions
there has been mention of default-value assignment when no 'real' value
had been found for a feature at the time of expression. With the (16") ver-

24 DikBakker
sion of the clause node above this is still the necessary procedure for Num-
ber in all VS cases, and for Gender in VXS cases. This mechanism should
receive more general attention since it is probably relevant for other mod-
ules in the FG model. An example is the insertion of the operator value
definite for terms that have a demonstrative operator like proximate or
remote in URs. Typically, but not necessarily, default values will be the
least marked values for some feature in a language, such as masculine and
singular for Arabic. To the extent that languages apply the same values for
such defaults, the mechanism should be treated at a more universal level of
the theory. For lack of such a default theory, and since the discussion
above focusses on one phenomenon in one language only, we will have to
supply the default values in an explicit and more or less ad hoc fashion
here. It will be assumed that in a more comprehensive fragment of the
grammar they will be provided by a general procedure which operates in-
dependently in all relevant instances. For the time being, (16"') will be an
updated version of the clause node for Arabic.
(16" ')Node k (uninstantiated)
Slot: clause
Cat: s
FncFtrs: TENSE, PERSON.subject < oo, default=3 >,
GENDER.subject < 1, default=m >,
NUMBER.subject < 0, default=sg >
A second point that should be made here concerns constituency. It could be
argued that in a more or less indirect manner constituency in our dynamic
expression rule component is defined by the templates, or rather the fillers
for the SubCat fields of the respective nodes. So, in the case of the clause
node in (18) above, we can say that the Cat fields of the fillers of the PI,
vfin and focus slots are the constituents of the s category of its own Cat.
For a language, the complete set of possible fillers of all Cat nodes dy-
namically and recursively defines the constituency of that language. It is
crucial, however, that the dynamics of the tree expansion process are in
tune with any static interpretation of constituency. More concretely, for the
Arabic data above we should check whether the breadth-first strategy has
implications for the way in which we represent compound noun phrases.
For the computation of the right form for (25a-b) - repeated below for
convenience - it seems to be attractive to assume that both conjuncts are

constituents at the clause level in their own right, expressed via serialized
slots, rather than as two subconstituents of a higher conjunct node. Finding
the right value for the Gender feature would then simply be taking the
value of the first constituent that comes in, i.e. masculine for (25a) and
feminine for (25b).
(25) a. Qara?a ?al-?awladu w-al-banatu kitaban.
read.3SG.M the-boys and-the-girls book
b. QaraPat l-banatu w-al-?awladu kitaban.
read.3SG.F the-girls and-the-boys book
However, such a 'flat' structure seems to make it problematic to cater for
the agreement in example (29) below where plurality is a feature of the
combined terms rather than the individual ones. Singular agreement is un-
grammatical here.
(29) Moroccan Arabic (Corbett 2000:202)
?umar w-?ali msaw/*msa.
Omar and Ali left.PL/left.SG.M
Omar and Ali left.'
Although proposals have been made for coordination in FG, there exists no
canonical underlying representation for coordinated terms, nor for coordi-
nates at any other level of grammatical description.23
It is therefore some-
what speculative to discuss them here, let alone to introduce a formal struc-
ture for their expression. What follows is therefore at best tentative.
The most straightforward way to represent the compound noun phrase
the girl and the boy seems to be the term structure in (30), which contains
the coordination operator '&'.
(30) ((def sg Xl : girl (x,)) & (def sg x2 : boy (x2 )))
A disadvantage of this representation may be the fact that its plurality in
the syntax should be inferred by the expression rules since it is not coded
in the underlying representation in a direct way. We also need, however,
the sg operators in order to get the right forms for the respective conjuncts.

26 Dik Bakker
An alternative would be to embed the conjuncts under a common Number
operator, as in (31), with its own variable.
(31) (pi x3: ((def sg x, : girl (X l )) & (def sg x2 : boy (x2 ))))
The assumption underlying (31) is that coordination creates a new referent
in the discourse, which is the union of the referents of the conjunct terms.
An extra complication for (31) is the question whether we also need to
insert the other term operators for the x3 variable, such as definiteness, and
what we would do in that respect with combinations like the one in (32).
(32) That boy and a girl entered the store.
For this discussion I will assume that (30) is at least a good candidate for
the representation of compound nouns. It has the advantage of simplicity,
but leaves part of the work to the expression rules. However, this would
not be a novelty altogether. For Gender the eventual determination of the
value for a compound term has to be left to the expression rules anyway,
since this feature is not coded as a term operator in the first place. Interest-
ingly, in Arabic there are other examples of the interaction between gender
and constituent order. A way of making yes-no questions is to prefix the
sentence with a form of the 3rd
person pronoun. According to Olmsted
Gary & Gamal-Eldin (1982) the number and gender of this prefixed pro-
noun are based on those of the subject. In the case of a coordinated subject,
the gender of the first conjunct is chosen, analogous to the situation with
postverbal subject agreement discussed above. This is shown in (33) be-
low. (33a) has a masculine first conjunct (proper name Sami) and (33b) a
feminine one (proper name Muna).
(33) Arabic (Olmsted Gary & Gamal-Eldin 1982:5)
a. Huwwa Sami wi Muna maguush?
3.M.SG Sami.M and Mona.F came.not.3PL
'Haven't Sami and Mona come?'
b. Hijja Muna wi Sami maguush?
3.F.SG Mona.F and Sami.M came.not.3PL
'Haven't Mona and Sami come?'

In general, therefore, resolution seems to be a rather natural part of (dy-
namic) expression rather than (static) semantic representation. As for the
corresponding underlying representation, we would like to claim as little
semantic structure as is strictly necessary.24
A more complex term structure
such as the one in (31) could then be claimed for cases where there is both
semantic and formal evidence for it. This is most probably the case for
examples such as the ones in (34) below, where the conjuncts are quanti-
fied by zuuz, 'both' and plural agreement is also obligatory for VS orders.
(34) Arabic (Aoun et al. 1994:211)
a. ?umar w Sa?id msaw b-zuuz
Omar and Said went.3PL with-both
b. *Msa ?umar w Sa?id b-zuuz
went.3SG Omar and Said with-both
Omar and Said both went to school'
For such utterances I would postulate the more complex term representa-
tion in (31). The binding variable x3 and its plural operator are presumably
introduced by the constituent containing the zuuz quantifier, which is
probably best analyzed as a satellite term at the level of the core predica-
tion. This term is in focus, therefore highly accessible and always ex-
pressed in the postverbal focus position. It could be said to bind both con-
juncts via its term variable.25
Against this background, verb agreement with compound terms in Ara-
bic may then be determined in the following way. Person is a feature of the
whole compound term. It will get value 3 if there is no [+S] or [+A] among
the conjuncts of the term. As a globally accessible feature it is established
at the sentence node and is available by inheritance. Both Gender and
Number are derived in a more dynamic sense from the subject term, when
it is processed itself. Both feature values will be established on the basis of
the values of the respective conjuncts. In a left-to-right manner, the overall
value will be updated under the application of the relevant resolution
schema. For Number, the initial value will be established as the value of
the first conjunct. If any conjuncts follow, this value will be overridden by
plural. In Arabic this process only has effect on agreement in SV cases,
when it reaches the vfin node via percolation and inheritance. It will be
assumed that Number cannot be retrieved from terms that are still being
processed in the background in parallel with the expression of an earlier
l-l-medrasa.
to-the-school
l-l-medrasa.
to-the-school

28 DikBakker
node, in other words it is inaccessible. Therefore all VS cases will get de-
fault singular agreement. For Gender, on the other hand, it works rather
differently. The initial value for Gender is also established as the gender of
the first conjunct term. As the conjuncts are processed and expressed se-
quentially their gender value will be taken into consideration by the resolu-
tion rule for [-S,-A] terms. In the case of Arabic this means that masculine
will override feminine. As a result the value for terms like the one in (30)
will end up being masculine. This works for SV cases. For VS constella-
tions we have already assumed above that Gender as opposed to Number is
accessible in the background process that runs in parallel with the Vfm
expression process. However, under the 'flat' term constellation proposed
earlier only gender information from the first conjunct of a compound term
in the background will be available in time for it to be taken into considera-
tion for agreement purposes. This explains why only the gender of the first
conjunct will be coded in the agreement marker on the verb.
5. Discussion and conclusions
On the basis of data from subject verb agreement in Standard Arabic it was
argued above that the five principles of dynamic expression proposed on
earlier occasions do not provide the complete framework necessary. In
order to explain why there is only a certain amount of gender agreement
with postverbal subjects rather than full agreement or none at all it was
assumed that during the expression of some underlying constituent there is
access to a parallel process that selects the next constituent. Only a subset
of the features of this constituent becomes available, i.e. the more accessi-
ble ones. These may be selected on the basis of their more prominent role
in reference tracking, or for other functionally motivated reasons.26
For
Arabic it appears that Gender is more outstanding than Number, at least in
the context of subject verb agreement. Access is also limited in another
sense viz. only the constituent selected to directly follow the one under
expression is available; in the case of compound constituents this concerns
only the leftmost conjunct. In the dynamic expression rule model and its
treelike representations this parallelism may be implemented in terms of a
restricted amount of breadth-first node development ('look ahead') coming
before any expansion in depth.
A slightly different way of looking at the role of parallelism, which in
my opinion would not affect the position taken here, is to see it as an ex-

planatory device in a diachronic sense rather than a mechanism which is
active in a synchronic grammar of Arabic. In other words, gender agree-
ment with the first conjunct would then be a grammaticalized result of this
parallellism and simply a rule applied by speakers of today's Arabic, who
learned it and internalized it at the language acquisition stage. Corbett
(2000:208) assumes that this may well be the case in Modern Standard
Arabic. Other varieties, including Moroccan Arabic according to Aoun et
al. (1996:196), have grammaticalized plural agreement in both orders, i.e.
there is no difference between the two situations in terms of agreement
marking. In yet other varieties, and notably Cairene Arabic, there actually
is variation in plural versus singular marking for 3rd
person referents even
for SV patterns. The crucial factor appears to be the score on the animacy
scale for the subject. On the basis of recorded interviews with 26 speakers
of Cairene Arabic Belnap (1999:174) observes the following for plural
preverbal subjects (my table is an aggregation of Belnap's table 1).
Table 1. Animacy and agreement in Cairene Arabic
Semantics of Number of instances Percentage plural
head noun in corpus marking on V
human 174 91%
animal 20 35%
inanimate 335 3%
This variation in Cairene Arabic may be indicative of the dynamic role of
the animacy parameter rather than its being grammaticalized completely
into fixed rules for this variety. The respective varieties of Arabic (and
other languages) could then be placed on the following number agreement
continuum. Languages to the far left have grammaticalized number agree-
ment for 3rd
person subjects in VS orders to singular, the languages to the
far right to plural, while the languages in the middle determine it on the
basis of its accessibility in the discourse situation. Animateness and Focal-
ity are factors which co-determine accessibility.
ISG - variation - PL |
Standard Cairene Moroccan
Arabic Arabic Arabic
Figure 3. Grammaticalization cline for 3rd person number agreement

30 Oik Bakker
But even if we assumed full grammaticalization for this agreement phe-
nomenon in Standard Arabic, we would still be left with the problem of
how to code this in the FG model. Even under the full grammaticalization
scenario we need a way to inspect the order in which conjunts will be ex-
pressed before the agreement parameters are determined. Of course, we
might make the '&' operator order-sensitive, just like the restrictor opera-
tor ':'. Or we might assign a pragmatic function to the first conjunct in
order to code its 'prominence'. Agreement could then be made sensitive to
such devices, and look ahead could be disposed of. We would, however,
still need a processing approach since they apply in VS constellations but
not in SV ones. And a look-ahead device or something similar is still nec-
essary for independent reasons i.e. if we want our model to explain why
these types of grammatical patterns have come about in the first place. We
also need look ahead if we want to explain why certain speech errors such
as so-called Spoonerisms are much more common than others.
Of course these observations have been based on a limited amount of
data, stemming mainly from one language only. It is unlikely, however,
that speaker models of other languages would work in a completely differ-
ent way as far as their dynamic operating principles are concerned. In this
respect it is interesting that form-order interactions are certainly not un-
common in the languages of the world, as shown in examples (1) - (4).
This means that a look-ahead mechanism might well turn out to be part of
the speaker model irrespective of the language, and that we could predict
that traces of it may be found in all spoken varieties, and not only in speech
errors. In details, however, languages may of course vary considerably. For
instance, the three agreement parameters Person, Number and Gender may
play a different role in different grammars and there may be different win-
dow sizes for different stages of the expression process. For Standard Ara-
bic the following accessibility hierarchy seems to apply, at least to subject
verb agreement:
(35) Person > Gender > Number
The hierarchy in (35) seems to be only partially in tune with some typo-
logical observations that were made concerning the synchronic and dia-
chronic behaviour of these three parameters in subject-verb agreement
(Siewierska & Bakker, fc.). In our sample of 426 languages of the world
subject-verb agreement is relevant in some 78% of the cases. In over 99%
of these languages Person is one of the agreement features, which points

towards its prominence over the other features. In 94% of these cases
Speaker and Addressee are formally contrasted while non-discourse par-
ticipants (i.e. 3rd
persons) are often homophonous with one of these forms,
or unmarked (zero). This confirms the prominent status of Speaker and
Addressee; the frequency of zero expression confirms the status of 3rd
per-
son as an 'else' category or a default. Diachronically, going from free pro-
nouns to verbal affixes, Person is almost never lost altogether. For the two
other features the picture is a bit different. Overall, Number plays a role in
subject agreement in 93% of the relevant languages, while in only 35%
Gender is manifest. The same relative distribution was found for the se-
mantic parameters of free pronominal forms, although the absolute num-
bers are higher. Furthermore, when Number gets lost from subject agree-
ment markers this is typically only partial (e.g. dual gets lost but not
plural), while Gender typically gets lost completely. It remains to be seen
whether a conclusion should be that the relationship between the overall
role of Number and Gender in grammars may differ from their role in the
dynamics of subject agreement. It may also be the case that Arabic deviates
from the mainstream in this respect in the sense that Gender plays a more
prominent role in this language than in many other languages. This view
may get some support from the fact that Arabic has gender not only in the
third person but also in the second person pronouns and in subject agree-
ment markers, which is relatively rare in our database. Gender is manifest
in both free pronominal forms and bound ones in only 16 languages, or less
than 5% of the languages that have subject agreement in the first place.
The value of a model of grammar is determined to a high degree on the
basis of the predictions that it makes, because these render the theory veri-
fiable and falsifiable in terms of real language data. Whatever the merits of
the expression scheme presented above, a number of predictions might be
derived from it concerning subject marking which, all other things being
equal, would potentially be different for languages of varying constituent
order types, for both their basic and marked orders. The same holds for
object agreement marking. It is not clear what types of predictions may be
derived on the basis of the standard model of expression in FG.
A final remark concerns underlying representations in Functional
Grammar. According to the theory, the criteria on the basis of which the
precise contents of UR's are determined are twofold. They should be pre-
cise enough to get at the right interpretation of the utterance in the dis-
course and to trigger the right expression rules. As in the case of the repre-
sentation of compound terms above, the latter requirement may well work

32 Dik Bakker
as a constraining factor on the former. Indeed, it is not always clear how to
determine what belongs to the pure semantics and pragmatics of a sentence
on the one hand and which implications and inferences might be derived
from them in the context on the other hand. Not subjecting oneself to the
exercise of working out the admittedly tedious step-by-step process that
eventually leads to the final form may both result in the unnecessary intro-
duction of some elements and the neglect of others. Therefore, even if the
priority in FG obviously lies with the functional side of language, work on
the more mundane aspects of grammar, i.e. morphosyntax, may well pro-
vide a better insight into the more invisible conceptual aspects which are
often invoked in order to explain the visible side of linguistic behaviour.
There is some risk involved, however, in that we may find that expression
has its own principles which cannot be reduced to functional explanation.
Abbreviations
0 null morpheme F feminine gender
1 first person FOC focus
2 second person LOC locative
3 third person Μ masculine gender
A addressee PAT patient
ABS absolutive PI person indicator
AG agent PL plural
AUX auxiliary POSS possessive
COLL collective PRF perfective
COMP complementizer PST past tense
DAT dative s speaker
DECL declarative SG singular
DEF definite TOP topic
DU dual TRANS transitive
ERG ergative

Notes
1. This was observed long ago. Corbett (1983:74) concludes that 'the further a
target is from its controller, the more frequently semantic agreement will oc-
cur [rather than grammatical agreement - DB]'.
2. Of course in the formal sense there is no problem if we assume that all order
phenomena can be predicted on the basis of functional information. But even
if this were the case, the cost would be to prerun the order rules before form is
calculated, i.e. by reversing the order of the two expression modules. This
would in turn endanger any case where order is determined by form. Dik
(1997:341) seems to acknowledge the form-order problem by giving an ex-
ample from Dutch, cited in (i) below.
(i) a. Jij zie-t een kat in de tuin.
2SG see-2SG a cat in the garden
'You see a cat in the garden.'
b. In de tuin zie je een kat.
In the garden see-0 2SG a cat
'In the garden you see a cat.'
In (ia), with a preverbal second person singular subject, there is an agreement
marker on the verb; in (ib), with a postverbal subject, there is no such marker.
No solution is provided for such constellations, however, by Dik (1997). Note
that a solution based on phonological reasoning, as suggested by an anonym-
nous reviewer, makes matters in fact maximally complex since the decision
about the form should then be postponed until the very end of the expression
process: the determination of the phonological form. In that case we would be
dealing with a deletion, a device not desired for the expression component.
Apart from that, there would be complications for such phonologically related
utterances as in (ii).
(ii) Je/Hij maak-t je broek vuil.
2SG/3SG make-2SG your pants dirty
'You are/He is dirtying your pants.'
If we base the phenomenon of (i) on phonological diachrony rather than an
active phonological rule, then again we do not have a solution, since a dia-
chronic rule can have no place in a synchronic grammar. In that case it should
be treated as a completely arbitrary rule in the sense of Croft (1995), with or-
der as the only 'explanatory' factor.
3. This is not to say, of course, that iconicity is not one of the more important
factors that determine linguistic form in general. As such, it is frequently and

34 Dik Bakker
convincingly invoked in functional and typological linguistic work. See Hai-
man (1983) for a first discussion; see Croft (1990: 164f) for an overview; and
see Hengeveld (1989) for an early example in FG, motivating the hierarchical
structure of underlying representations. However, the complex and varied
ways in which the different types of primary and auxiliary operators project
onto form across languages calls for a more comprehensive approach, involv-
ing multiple explanatory principles. Also, even if iconicity is assigned a very
central explanatory role, as is done in the case of the centripetal principle in
Dik (1997), it may be assumed to have shaped the expression rules together
with other grammar external principles, such as economy and processing, and
that they are still active in continuously reshaping them. This still leaves the
theory with the task of providing the precise rules for getting at the right syn-
tactic and morphological elements in the synchronic grammar of some lan-
guage.
4. This is more or less suggested by the way the so-called contextual retrieval
principle is defined. In order to assemble the necessary primary operators for
some expression rule 'the context in which it occurs in [the] underlying clause
structure is searched for relevant agreement features' (Dik 1997:356). Before
expression takes place the underlying representation is then extended pre-
cisely such that copies of relevant operators are locally available. It is not
clear from this how the searching process would take place other than taking
precisely the set of operators that are necessary for the job on hindsight,
thereby becoming circular. In a way, this would locate aspects of expression
in the semantics of the theory, and could be interpreted as 'syntactic smug-
gling', as a correlate to the 'semantic smuggling' which is sometimes ob-
served in the practice of formal syntacticians, who turn [+animate] or [+topic]
into syntactic features.
5. This is a mild form of autonomy of morphosyntax which following Croft
(1995) may be called 'arbitrairiness'. The 2SG phenomenon of Dutch as dis-
cussed in note 2 may be an example of this. A stronger form of assigning
autonomy to the formal component of the grammar is assuming that there are
inherent ('innate') principles at work, such as those of Chomsky's Universal
Grammar. One could reserve the notion 'autonomy' for this type of independ-
ence of morpho-syntax. Our five principles and the constraints they create for
expression may be seen as aspects of such autonomy.
6. The notion 'cognitive adequacy' replaces the traditional notion of psycholin-
guistic adequacy.
7. Currently, authors in the field seem to disagree as to whether two more or less
independent models should be distinguished or only one. Brown & Hagoort
(1999) devote two separate chapters to the two modes of language processing,

written by different authors. Others argue that the overlap between the two
modalities is such - the speaker is also an addressee and the other way around
- that an integrated model is called for (Gerard Kempen p.c.). This does not
affect our argument for a cognitively adequate model, however.
8. In the formalizations below, upper case notions indicate feature names
(TENSE), and lower case notions indicate feature values (past). Quotes indi-
cate predicates of the object language ('girl'). When there may be ambiguity
among feature names, they are made more specific by way of an extension:
e.g. NUMBER.subject means the Number value of the Subject term. All these
notations are relatively informal, and meant only to illustrate the points under
discussion. A formally more rigourous version, based on feature-value logic
and unification, has been introduced in Bakker (1989). A computer implemen-
tation is discussed in Bakker (1994).
9. In the domain of phonology backward vowel harmony would be an example.
10. Mohammad (2000) discusses and contrasts a number of Arabic dialects.
Unless stated otherwise I will only use examples here which are labelled as
Standard Arabic by this author.
11. See Siewierska (fc.) for cross-linguistic differences between first and second
person versus 3rd
person forms.
12. The three labels Person, Number and Gender are not uniquely determinable at
the clause level since there may be more terms which have operator values
that satisfy them. They therefore remain uninstantiated.
13. Such lexical priority rules are also necessary for the insertion of irregular
forms during the expression process. Their role in expression is discussed in
somewhat more detail in Siewierska & Bakker (this volume).
14. For this discussion I will assume that a monitor will keep track of the contents
of the Config, will mark all material that has been expressed, and will send a
signal when there is no unmarked material left. An alternative way would be
to 'do away' with material that has been expressed. This would be more in
keeping with the literal meaning of the notion of 'expression', but it would fail
for cases where material is expressed more than once. Furthermore, there is
still the Focus slot to be filled. The problem of how to deal with empty slots,
and whether they should be generated in the first place, will not be discussed
here. See Bakker (2001) for some remarks on empty slots.
15. Some authors in the generative literature, and notably Mohammad (2000) and
van Gelderen (1996), argue that the VS order should be interpreted as a covert
expletive, which explains for the 3rd
person singular agreement. Although a
case for a 'pro-dropped' expletive could technically also be made within an
FG approach, this does not seem to be a suitable solution for the gender
agreement phenomena to be discussed below, which are in need of a funda-

36 DikBakker
mentally different treatment in my view. For lack of an explicit expletive pro
form I would prefer to interpret the third person singular verb form in these
cases as a somewhat underrepresented version of 'the subject is a non-speech
participant'. Aoun et al. (1994:200f), who take a generative perspective, reject
an analysis based on expletives on theory-internal grounds.
16. For an idealized model of the speaker, who makes no speech errors, I will
assume that harmony between inherited and local features is always the case.
However, for a more realistic model of the speaker, and for one of the ad-
dressee, there should of course be provisions for the recognition (and crea-
tion!) of ill-formed utterances.
17. It may look rather arbitrary to postulate this for the respective features con-
cerned. However, apart from the fact that the empirical data point in the direc-
tion of Number being less prominent than Person and Gender in these utter-
ances of Arabic, it is known from work in typology that Person is the most
prominent feature in verb agreement markers and pronouns in the languages
of the world. Its major function is the identification of speech participants, i.e.
1st
and 2nd
person, versus non-participants, or 3rd
person. The major function
of Gender and Number seems to be the identification of discourse referents,
which is mainly a subdistinction of the currently relevant 3rd
person referents.
Cf. Bhatt (to appear) for an in-depth discussion of this.
18. These examples seem to support the 'underrepresented 3rd person subject'
interpretation that I suggested above in note 15 as opposed to the expletive in-
terpretation of some authors.
19. As a technical term, look ahead was introduced in the theory of parsing natu-
ral language rather than the generation of sentences, as is the case here (cf.
van der Steen 1987). It remains to be seen whether in a model that represents
both the Speaker and the Addressee it may in fact turn out to be one and the
same notion in both modalities.
20. For this discussion it suffices to assume that the upper limit for global acces-
sibility is the information contained in the underlying clause (but see Siewier-
ska & Bakker (this volume) for a discourse approach to agreement and cross-
reference phenomena). The notion 'accessibility' is used as an English word
rather than a technical term. While it is interpreted here from the perspective
of a model of the Speaker, other authors - most notably Ariel (2000) - use it
from the perspective of a model of the Addressee.
21. Undoubtedly there will be lower limits also to such a process. For a discussion
of this point see Bakker (1994:283ff). For the current discussion it will be as-
sumed that any higher level of underlying representations that is recursively
invoked within a lower level creates such a barrier.

22. Technically, we are forced to make a choice between depth first and breadth
first. The latter seems to win out in the case at hand. However, since in the
model we are dealing with the mapping of a no doubt very complex parallel
process onto the much simpler dynamic model, this is only an either/or choice
on first sight. In fact we have a mixed situation here, in which partial breadth-
first development alternates with extension in depth.
23. This is extra remarkable since a functional theory of grammar was first sug-
gested in Simon Dik's (1968) PhD thesis entitled 'Coordination: Its implica-
tions for the theory of general linguistics'.
24. Aoun et al. (1994:207), who try to explain the agreement phenomena of Ara-
bic discussed above in terms of structural relationships such as specifier-head
and government, represent what they call the prominence of the first conjunct
in postverbal coordinated structures by putting the first element in the speci-
fier position of the NP where it governs the other conjuncts. This could be
seen as a static version of a 'look-ahead' mechanism. Note that this represen-
tation is only assumed for VS constellations; in other words constituent order
is relevant in these cases, as admitted by Aoun et al. (1994:219).
25. My expectation would be that plural would also emerge in VS orders under
the presence of other types of first-order satellites which predicate something
about the subject rather than about the verbal predicate, such as 'sick' or 'with
pleasure'. A representation as in (31) could also be claimed for collective as
opposed to distributive readings, and for predicates that force a collective
reading onto their arguments, such as intransitive 'meet'.
26. This may be true of information processing in general. In the minimalist
framework (Chomsky 1992:44) a distinction is made between strong and
weak inflection, and correspondingly between strong and weak NP-features,
precisely to cater for agreement differences due to constituent order, as in
Arabic. This may be the formal translation of what I consider to be function-
ally determined agreement phenomena.
References
Aoun, Joseph, Elabbas Bennamoun and Dominique Sportiche
1994 Agreement, word order, and conjunction in some varieties of Arabic.
Linguistic Inquiry 25-2: 195-220.
Ariel, Mira
2000 The development of person agreement markers: from pronouns to
higher accessibility markers. In Usage-based models of language,

38 DikBakker
Michael Barlow and Suzanne Kemmer (eds.), 197-260. Stanford:
CSLI Publications.
Bahloul, Maher and Wayne Harbert
1993 Agreement Asymmetries in Arabic. In The proceedings of the elev-
enth West Coast conference on formal linguistics, Jonathan Mead
(ed.), 15-31. Los Angeles: University of California Press.
Bakker, Dik
1989 A formalism for Functional Grammar expression rules. In Func-
tional Grammar and the computer, John H. Connolly and Simon C.
Dik (eds.), 45-64. Dordrecht: Foris.
1994 Formal and computational aspects of Functional Grammar and
language typology. Amsterdam: IFOTT.
1999 FG expression rules: from templates to constituent structure. Work-
ing papers in Functional Grammar 67, University of Amsterdam.
2001 The FG expression rules: a dynamic model. Revista Canaria de
Estudios Ingleses 42: 15-54.
Bakker, Dik and Anna Siewierska
2002 Adpositions, the Lexicon and Expression Rules. In New perspectives
on argument structure in Functional Grammar, Ricardo Mairal
Uson and Maria J. Perez Quintero (eds.), 125-178. Berlin: Mouton
de Gruyter.
Belnap, Robert K.
1999 A new perspective on the history of Arabic variation in marking
agreement with plural heads. Folia Linguistica XXXIII/2: 169-185.
Bhatt, D.N.S.
fc. Pronouns: a crosslinguistic study. Oxford: Oxford University Press.
Brown, Colin M. and Peter Hagoort (eds.)
1999 The neurocognition of language. Oxford: Oxford University Press.
Chomsky, Noam
1992 A Minimalist Program for Linguistic Theory. MIT Occasional Pa-
pers in Linguistics. Cambridge: MIT Press.
Corbett, Greville G.
1983 Hierarchies. Targets and controllers: Agreement patterns in Slavic.
London: Croom Helm.
1991 Gender. Cambridge: Cambridge University Press.
2000 Number. Cambridge: Cambridge University Press.
Croft, William C.
1990 Typology and universals. Cambridge: Cambridge University Press.
1995 Autonomy and functionalist linguistics. Language 71-3: 490-532.

Dik, Simon C.
1997 The Theory of Functional Grammar 1: The structure of the clause.
2nd
, revised edition, edited by K. Hengeveld. Berlin: Mouton de
Gruyter.
Foley, William A.
1986 The Papuan languages of New Guinea. Cambridge: Cambridge
University Press.
1991 The Yimas Language of New Guinea. Stanford: Stanford University
Press.
Friberg, Barbara
1991 Konjo's peripatetic person markers. In Papers in Austronesian lin-
guistics, Hein Steinhauer (ed.), 137-171. Pacific Linguistics A84.
Gelderen, Elly van
1996 Parametrizing agreement features in Arabic, Bantu languages, and
varieties of English. Linguistics 34: 753-767.
Haegeman, Liliane M.
1994 Introduction to government and binding theory. Oxford: Basil
Blackwell.
Haiman, John
1980 Hua: a Papuan language of the Eastern Highlands of New Guinea.
Amsterdam: John Benjamins.
1983 Iconic and economic motivation. Language 54: 781-819.
Hengeveld, Kees
1989 Layers and operators in Functional Grammar. Journal of Linguistics
25.1: 127-157.
2004 The architecture of a Functional Discourse Grammar. In A new ar-
chitecture for Functional Grammar, J.Lachlan Mackenzie and Maria
A. Gomez Gonzalez (eds.). Berlin: Mouton de Gruyter.
Hieda, Osamu
1998 A sketch of Koegu grammar. In Surmic languages and cultures,
Gerrit J. Dimmendaal and Marco Last (eds.), 345-373. Köln: Rüdi-
ger Köppe Verlag.
Mohammad, Mohammad A.
2000 Word order, agreement and pronominalization in standard and
Palestinian Arabic. Amsterdam: John Benjamins.
Najlis, Elena
1973 Lengua Selknam. Filologiay lingiiistica. Buenos Aires: Universidad
del Salvador.

Exploring the Variety of Random
Documents with Different Content

he had passed a lonely little Saeter. He had only a vague idea of his
bearings; and, indeed, without knowing it, as he tried to retrace his
steps he was wandering farther away, both from that Saeter and
from the Gaard.
He became distressed about the anxiety which his prolonged
absence would be causing to his friends: to dear old Knutty, who
had seen him start off so sadly: to his boy: to Katharine. He knew
that they were waiting for him, and wanting him, and that they were
watching the storm, and watching the evening fading into the night.
He knew so well that Knutty would pretend not to be troubled, and
would scold every one who even suggested that there might be
cause for anxiety. He almost heard her saying:
"He loves a thunderstorm. The silly fellow, I know him well!"
He smiled as he thought of her.
"My dear old Dane!" he said. "My dear old brick of a Dane!"
He wandered on and on trying to find the Saeter, changing his
direction several times, but in vain. But at last he caught sight of a
habitation at some distance, and made straight for it, thankful to
have found a haven. There was a light in the hut. Clifford knocked,
and the door was instantly opened. There was a fire in the stove.
"Ak," said the old woman who opened the door, "I thought it was
my son. But you are welcome. It is a fearful night. Many times I
thought the hut was struck. I am glad for company."
The son came in a few minutes afterwards, and she made hot
coffee for them both, whilst they dried themselves before the
crackling logs. And overcome by the genial warmth and his long
wanderings, Clifford slept.
And he dreamed of Katharine. He dreamed that he, who had
always found speech difficult, was able to tell her the story of
Marianne's death. He dreamed that he went on telling her, and she
went on listening; and it was such an easy matter to tell, that he
only wondered he had been silent so long.
"And that is all," he said, and he waited for her to speak as she
turned her dear face towards him. But when she was beginning to
speak, he awoke.

He awoke, glad and strong. He who had come out broken and
embittered, was going back made whole and sound. He thought of
his last words to Knutty:
"I shall be better later."
They had come true. The long wrestle with morbid
conscientiousness, his defeat, his wanderings, the great storm, the
safe arrival at a haven, his dream, and now his glad awakening had
made him whole.
The storm had died down about two in the morning, and it was
nearly six before he awoke. He could scarcely wait to drink the
coffee which the old woman prepared for him; scarcely wait to hear
her directions for getting back to the Gaard. He was off like some
impatient boy before she had finished telling him.
His step was brisk, his heart was light, his grave face was smiling.
He sang. He did not notice that the way was long and rough.
Everything in life seemed easy to him. He trod on air. At last, after
several hours, he saw the smoke of the Solli Gaard. He hastened
through the birch-woods, down the hillside, and into the courtyard.
There was a group of people standing round the carriage, which had
evidently just come back from a journey. Mor Inga and Gerda were
helping Knutty out of the carriage. Ejnar, Alan, and the Sorenskriver,
Solli, Ragnhild, and every one belonging to the Gaard, including old
Kari, crowded round her.
"Thank God, thank God, it was not he," she was saying.
Then old Kari looked up and saw Clifford. She firmly believed him
to be dead and thought this was his ghost.
"Aa Jösses!"[Q] she cried, falling down on her knees and folding
her hands in prayer.
They all turned and saw him. Alan rushed forward to meet his
father.
"Oh, father," the boy cried, "we thought you were dead—killed by
the lightning."
Then his pent-up feelings found their freedom in an outburst of
passionate, healing tears. Clifford folded him in his arms and
comforted him.

"And you cared so much?" the father asked, with a thrill of
gladness.
"Yes, yes!" the boy whispered, clinging close to him.
Then arm-in-arm they came to Knutty, who in her unselfishness
had stood back, wanting her two icebergs to have their meeting to
themselves.
"Dear one," she said, with tears in her eyes, "I have done all my
crying, and every one can tell you that I have behaved disgracefully.
And now I can do my scolding. How dared you give us so much
anxiety? Ak, it is all too much for me. I'm going to cry after all."
He stooped and kissed her hand.
"Don't scold me, and don't cry, dear Knutty," he said. "I have
come back from the mountains strong and glad."
They all pressed round him, greeting him warmly. Every one
belonging to the Gaard seemed to him to be there, except Katharine.
And he hungered for the sight of her.
"Knutty," he asked, "where is she—where is Miss Frensham?"
Knutty led him away and told him in broken words the history of
the morning, and their fearful anxiety, and Katharine's tender
kindness.
"And she stayed there with the dead Englishman," Knutty said
gently. "She said she could not leave him alone, and that you would
understand. She said you would come down safely from the
mountains, and the joy of reunion would be ours, and that she
would be with us in spirit. I know, kjaere, she suffered greatly in
staying behind."
The man's lip quivered.
"I will go to her," he said.
And the next moment he had prevailed on Solli to change the
horses and let Jens come with him. It was all done so quickly that
Solli had no time to relent. Clifford sprang in, signed to Alan to
follow him, and they were off. Old Kari, rather sullen at having been
done out of the ghost, retired crestfallen to the cowhouse.
But Gerda and Tante, Mor Inga and Ragnhild, stood watching the
carriage until it had wound round the hill and was out of sight.

"Nå," said Gerda, turning to Tante, "I begin to think that your
Englishman is going to fall in love with Fröken Frensham. Who would
have imagined such a thing?"
"Every one except you," replied Tante, giving her a hug.
"And why not myself?" asked Gerda.
"Because you are an unilluminated botanical duffer!" answered
Tante.
CHAPTER XVI.
Katharine lingered a little while longer at the Skyds-station to
comfort, by her sympathetic presence, the brother and friend of the
dead Englishman. To the end of their lives they remembered her
ministration. She gave out to them royally in generous fashion. It
was nothing to her that they were strangers; it was everything to
her that they were in trouble and needed a little human kindness.
They themselves had forgotten that they were strangers to her. It
was a pathetic tribute to her powers of sympathy that they both
spoke of the dead man as if she had known him.
"You remember," the brother said, "he never did care for fishing.
It always bored him, didn't it?"
"Yes," said Katharine gently.
"Do you remember him saying a few years ago," the friend said,
"that he should love to die on the mountains? He always loved the
mountains."
"Yes," said Katharine gently.
She scarcely had the heart to leave them; but at last she rose to
go, telling them there was an Englishman at the Solli Gaard who
spoke Norwegian well, and who would come to help them.
"He is the one for whom we came to seek here," she said, looking
away from them. "We are not yet sure that he is safe; but if he
comes down from the mountains, I know he will hasten to help you
about——"
They bowed their heads silently as she broke off.
"We shall take him home to England," the brother said.

"I am glad he will rest in his own country," Katharine answered.
The people of the Skyds-station fulfilled their promise to Solli, and
put Katharine in their best cariole. The two strangers helped her to
get in, and then stood watching her. They could not speak. But when
she held out her hand in farewell greeting, each man took it and
reverently kissed it. She was touched by their silent gratitude, and
the tears came into her eyes.
"I am so thankful I stayed behind," she said.
Then the driver, a little fellow of about twelve years old, whipped
up the yellow pony, and the Skyds-station was soon out of sight.
"And now, if indeed he has come back, I shall see him," Katharine
thought, with a thrill of happiness.
At the Skyds-station, when, by her own choice, she was left
alone, she had for the moment felt the bitterness of being outside
everything. She remembered her own words:
"He will come down from the mountains, and the joy of reunion
will be theirs, and I shall be outside of it—outside of it as always.
Always outside the heart of things."
That moment had been only one of the many times of passing
sadness and bitterness in Katharine's life, when she had said and felt
that she was outside everything: outside the inner heart of
friendship which never fails, outside ambitious achievement, outside
the region of great gifts, great talents, outside the magic world of
imagination, outside love. Her friend had died, her girlhood's lover
had died, her brother had failed her. She was alone, a solitary
spectator of other people's close friendships, passionate love,
successful work, absorbing careers; alone, outside the barrier which
separates all restless yearning spirits from that dim Land of Promise;
alone, outside. She, ever unconscious of her own genius of giving,
had no means of knowing that, by a mysterious dispensation, those
who give of themselves royally, without measure, are destined to go
out alone into the darkness of the night; alone, outside everything in
life.
But no such sad reflections came to Katharine now, as she sped
along the narrow valley, by the side of the glacier-river. Her thoughts
turned to Clifford and Knutty and Alan in loving unselfishness.

"The boy will have seen his dear father, and will now be
comforted," she said.
"Knutty will have seen her Englishman, and will now raise her old
head again," she said.
"Ah, how I hope and hope he was there to receive them when
they got back to the Gaard," she said.
"And now I shall see him, and the joy of reunion will be mine,"
she said.
But in the midst of her happy thoughts and yearnings, she did not
forget those two lonely compatriots and that silent companion in the
bedroom of the Skyds-station.
"My poor strangers," she said, "we will not forsake you."
They had come to the place where the sudden break in the valley
had cheered them during that terrible drive of the morning.
"Yes," thought Katharine, "that gave us hope this morning. I
should recognise this spot anywhere on earth. It was here I began
to have a strong belief that it could not be he lying dead at the
Skyds-station."
"Oh," she thought, with a shudder, "if it had been he!—if it had
been he!"
And her own words echoed back to her as an answer:
"My womanhood would be buried with my girlhood."
Then she looked up and saw a carriage in the distance, in the far
distance. The boy also saw it. As it approached nearer he said:
"It is from the Solli Gaard. That is Jens driving."
Katharine's heart gave a sudden bound.
"Haste, haste!" she said excitedly to the boy; and he, moved by
her eagerness, urged on the little yellow pony, which rose to the
occasion and flew over the ground.
Carriage and cariole drew up at the same moment, and Katharine
saw face to face the man whom she loved.
"We came to fetch you," he said.
CHAPTER XVII.

Bedstefar had been dead for three days, and it had been
arranged that the funeral was to take place a week after the night of
his death. Preparations had been going steadily forward, interrupted
only by the anxiety and excitement caused by Clifford's long absence
in the mountains and his supposed death. Bedstemor herself had
been much troubled about him, and had spent a good deal of time
watching for him. But when he returned safely, she felt free to
continue her persecutions in the kitchen; and it took a great amount
of Knutty's craftiness to entice her into the porch and keep her
there. Bedstemor was astonishingly well, seemed in excellent spirits,
and in answer to questions as to how she felt, she always said
briskly:
"Bra', bra,' meget bra'" (Well, very well).
Indeed, she was not a little gratified to be once more the central
figure of circumstances, as in the old days, before she and her
husband had retired to the dower-house. But, spite of her
cheerfulness, she looked a pathetic old figure wandering about,
relieved from constant attendance at her sick husband's bedside,
and thus thrown on the little outside world for distraction and
company. Tante was endlessly kind to her, but had many a secret
laugh over the old widow's unfunereal attitude of mind, and over her
stubborn determination to go and bully every one in the kitchen.
Tante herself was in great form again. She had recovered from her
fears and tears, and had, so she told Katharine, regained her usual
Viking bearing.
"Never shall I forget your tenderness, dear one," she said to
Katharine. "If I loved him even a hundred times more than I do, I
should not grudge him to you. He loves you, and you are the right
aura for him. And some day he will tell you so, although it will not be
very soon, stupid fellow! He will try and try many times, and leave
off suddenly. I know him well, my prisoner of silence. These
reserved people! What a nuisance they are to themselves, and every
one else! But to themselves—ak, ak, poor devils!"
Katharine, who was standing at the time on Knutty's bedroom-
balcony, looked out into the distance. She herself had been
somewhat silent since that sad morning at the Skyds-station.

"The end of it all will be, dear one," Knutty continued recklessly,
"that you will have to help him. This sort of man always has to be
helped, otherwise he goes on beginning and leaving off suddenly
until Doomsday. I know the genus well."
Katharine went away.
"Aha," said Knutty to herself, "I have said too much. And, after
all, it is premature. Oh, these parish-clocks! Why, Marianne has only
been dead about a year. How like her, only to have been dead about
a year! Oh, oh, what a wicked old woman I am!"
She called Katharine, and Katharine came.
"Kjaere," she said, as she stroked Katharine's hand lovingly. "I
have always been a free-lance with my naughty old tongue. No one
with any sense takes any notice of me. And am I not funny and
human too? All this time I have only been thinking that you are the
right aura for my Clifford. Not once have I asked myself whether my
Clifford were the right aura for you! I should have been an ideal
mother, always on the alert to snatch up all the best things for those
I loved, regardless of other people's feelings and interests. Ah, that
is right, you are smiling, and not angry with your Viking friend. And,
dear one, that reminds me again of how you comforted me when I
was not behaving like a Viking. Do you remember assuring me that
his absence, and Alan's anxiety for him, were working for their
complete reconciliation? Your words have come beautifully true,
haven't they? Well, you have the great heart that knows."
They were a small party at the Gaard now. Ejnar had gone off to
Kongsvold in the Dovre mountains, a district specially interesting to
botanists as the habitat of certain plants not found elsewhere. Gerda
would have gone with him, but that she had sprained her ankle. She
fretted for her Ejnar, although she pretended that his absence was a
great relief.
"It is grand to be free at last!" she said to Tante. "Free at last. I
can now take a long breath."
"Yes," said Tante, smiling mischievously, "freedom is delightful
when it does not make your nose red and your eyes moist!"
Alan had gone off with Jens to a mountain-lake to catch trout for
the funeral, and would not be back for a day or two; and Clifford

was away at the Skyds-station, helping the two strangers to make
the necessary arrangements for taking their sad burden home to
England. All the other guests except the Sorenskriver had left, and
he was in a thoroughly disagreeable mood, grumbling about the
food, and annoyed because there was going to be a funeral at the
Gaard.
"Then why not go away?" Katharine suggested on one occasion,
when his martyrdom had reached an acute stage.
"Thank you, I choose to stay," he answered in his gruffest tone of
voice.
Katharine laughed. She liked the Sorenskriver even at his worst.
"Read this German newspaper, with a whole column of abuse
against England," Katharine said, teasing him. "That will make you
feel cheerful, Sorenskriver."
"Sniksnak!" said the Sorenskriver, a little less roughly.
"Or come out for a walk with me and help me pick multebaer,"
she added. "Mor Inga was saying she had not half enough as yet."
"Perhaps I will come," he answered, with a grim smile on his face.
He took pleasure in Katharine's company, and was secretly delighted
that Clifford was busy helping those Englishmen over at the Skyds-
station. In this way he got Katharine to himself, and he sat smoking
his long pipe in the porch, grumbling and disagreeable, but, in
justice it must be owned, gentle to Bedstemor. Tante declared that
he was courting Katharine.
"I am given to understand, dear one," she said, with a twinkle in
her eye, "that the Norwegian way of courting is to be extremely
disagreeable, and almost rude to the person whom you adore. In a
day or two you will have a proposal—and what then?"
"Tante thinks only about marriages," Gerda said reproachfully.
"Well, what else in the world is there to think about?" Tante asked
defiantly.
"Oh, Tante, you know you do not think that," Gerda said. "If you
really thought that, why didn't you get married yourself?"
"Because, kjaere, no one would have me, except a sea-captain,
and he was mad," Knutty answered. "And he killed his mate soon
afterwards. I was always glad I was not his mate!"

"It is not true," Gerda said, turning indignantly to Katharine. "She
had lots of admirers and lovers. You ask her Englishman. He knows."
"Ah," said Knutty, "perhaps I did have a few admirers in my time!
You may be sure no sane woman would ever say she had never had
any, unless there was some one at hand to deny her statement."
When Clifford came home that evening, Knutty herself broached
the subject again.
"Kjaere," she said, "did I have a few admirers in my time, or did I
not? I have forgotten. Not that a woman ever does forget, but tell
me!"
"You had numbers, Knutty," Clifford answered, smiling at her;
"and I was jealous of them all. At nine I was jealous of the sea-
captain, and at ten I was jealous of the clergyman in Jutland, and at
twelve of the English architect, and at thirteen of the Swedish officer,
and so on and so on."
Later in the evening, when he and Katharine were sitting alone
near the great hay-barn, Katharine spoke of Knutty.
"She is the dearest old woman I have ever met," she said warmly.
"I don't wonder that you all love her."
"I can never tell you what she has been to me," he answered. "It
was always a great grief to me that——" He broke off.
"It was a great grief to me that——"
Again he broke off. He was trying to speak of Knutty's indifference
to Marianne; and even this was too hard for him to say. Up in the
mountains, he had felt that it would be easy for him to tell Katharine
everything that he had in his heart, beginning with the story of
Marianne and Marianne's death, and ending with himself and his
love for her. But now that he was near her, he could say nothing
about his own personal life and inner feelings. He could only bend
forward and scratch a hole in the ground with his stick. Katharine
remembered how Knutty had spoken of his "beginnings" and
"breakings off," and she said:
"Knutty understands you through and through, Professor
Thornton. Doesn't she?"
"Yes," he answered simply. "But why should you say that just
now?"

"Oh, I don't know," Katharine answered. "I was thinking of her,
and it came into my head. And I was so touched by her grief when
she feared that she—we—had lost you."
"I do not know what she and the boy would have done without
you," he said, still working with that stick.
Katharine was silent.
"And I cannot think what those men over at the Skyds-station
would have done without you," he said. "Their last words to me this
afternoon were, 'Tell her we shall always be wishing to serve her.'"
Katharine remained silent.
"There was this little packet which I was to give you," he said,
after a pause. "It was the poor fellow's South African service-medal.
You were to have it."
He watched her as she opened the packet and touched the
medal. He watched her as she put it in the palm of her hand and
looked at it with dim eyes. It would have been easy for him to have
opened his heart to her then and there, if he could only have known
that she was saying to him with speechless tongue:
"My own dear love, whilst I am looking at this soldier's medal, my
heart is giving thanks that the lightning spared you to me."
But he could not guess that, and the moment passed.
The next day, when they were again alone, he attempted to
speak.
"Do you remember my saying up at the Saeter that I tried never
to dream?" he began.
"Yes," she said. "I have always wished to ask you why you should
feel so strongly about dreams."
"I should like to tell you," he said eagerly. "I want to tell you. But
——"
He broke off again, and turned to her with a pathetic smile on his
face.
"Speech has never been easy to me," he said.
CHAPTER XVIII.
BEDSTEFAR'S FUNERAL.

The day before Bedstefar's funeral Jens and Alan came down
from the mountain-lake laden with nearly two hundred pounds of
trout, and the cotters' children finished their task of bringing in all
the multebaer they could find; for no Norwegian entertainment,
taking place at this season of the year, would have been considered
complete without this much-loved fruit; and certainly it would seem
that multebaer had a softening effect on the strange and somewhat
hard Norwegian temperament. As Tante said, from her own personal
observations of the previous days, multebaer spelt magic!
"Ibsen has not done justice to his country," she told Gerda. "He
ought at least once to have described them as being under the
influence of these berries. Then a softer side of their nature would
have been made apparent to all. Why, the Sorenskriver himself
becomes a woolly lamb as he bends over his plate of cloudberries-
and-cream. He ought to have his photograph taken. No one would
recognise him, and that is what photographs are for!"
They all helped to decorate the Gaard inside and out with
branches of firs and birches. Bedstefar's black house was decorated
too, and the whole courtyard was covered with sprigs of juniper and
fir. A beautiful arch of fir and birch was raised over the white gate
through which he would pass for the last time on his way down to
the old church in the valley.
Katharine, together with Ragnhild and Ingeborg, spent many
hours making strips of wreathing from twigs of the various berry-
shrubs up in the woods. Karl used these for lettering; so that
stretched from side to side of the arch ran the words, "Farvel, kjaere
Bedstefar."
When he had finished, every one came out to see his work, and
Mor Inga, turning to Tante, said proudly:
"My Karl is clever, isn't he?"
And she whispered:
"Three years ago he did that for our eldest son, and bitterly we
were weeping then. I go about thinking of that now."
Then Tante and Mor Inga took a little stroll away from the others,
outside the gate and down the road towards the great cowhouse.
Part of this road, too, had been planted with tall fir-branches, so that

Bedstefar would pass under the archway and through an avenue of
green until he reached the outer white gate, which was the entrance
to the Gaard enclosure. And here Mor Inga and Tante lingered,
whilst the proud Norwegian heart gave vent to its sadness, and the
kindly Danish heart beat in understanding sympathy, and the dead
son's dog Jeppe came and whined softly in token that he too was
mourning in remembrance of the past.
So the night, the bright Norwegian night, beginning to realise that
its brightness was being threatened, seeing that the birches were
counting their yellow leaves, even as we, no longer young and not
yet old, count our grey hairs, this summer night passed almost
imperceptibly into morning, and the activities of the next day began
early.
Bedstemor, reinstated in her former rôle of leading lady of the
Gaard, was in a state of feverish excitement. She was dressed in
black, and wore over her bodice a fine black silk shawl one hundred
years old. Her head was encased in a sort of black silk night-cap,
edged with old white lace: so that her pretty face was framed in
white. A slight flush on her cheeks made her look strangely youthful.
She sat in the porch waiting to receive the guests; and by special
request of Mor Inga and Solli himself, Tante, Gerda, and Katharine
sat there too. They felt awkward at first, knowing themselves to be
there in the capacity of sightseers rather than that of mourners; but
Bedstemor's cheerful spirits put them at their ease. She was much
interested in Katharine's dress-material, feeling the texture and
comparing it with her own.
"It is very good," she said thoughtfully, "but not so good as
mine!"
All the same, that dress-material worried her; she fingered it
several times, nodded mysteriously, and seemed lost in thought;
whether about Bedstefar or the dress-material, no one could of
course decide. But, later, she spoke of some wreaths which had
been sent, and she said quaintly:
"Min mand did not want any flowers. But it does not matter much
what he wanted. He won't know, stakkar, will he?"

At last the guests began to arrive, some in carioles, some in stol-
kjaerres, and some few in ordinary carriages. They all brought
funeral-cakes in large painted baskets. As each conveyance drove up
into the courtyard, one of the daughters, either Ragnhild, or
Ingeborg, or Helga, went out to meet it, greeted the guests, and
bore away the cake into the kitchen. It seemed to be the etiquette
that the cake should be received in person by one of the family. The
horses, most of them the knowing little Norwegian yellow Nordfjord
pony, or else the somewhat bigger Gudbrandsdal black horse, were
unharnessed and led away by the cotters. The guests advanced
awkwardly to the porch, greeted Bedstemor, and turned to the
strangers shyly, but were at once reassured by Tante's genial bearing
and Katharine's friendly smile. Gerda, too, was at her best, and was
feeling so cheerful that Tante feared she was going to break into
song. Quaint, strange-looking people crossed that threshold, shook
hands with every one in the porch, and passed into the house to find
Bedstemor, who had disappeared into the hall, and was seated in a
corner drinking port wine with an old friend. Wine and coffee were
served at once, as a sign of welcome to the Gaard. The flag, which
had been lowered to half-mast since Bedstefar's death, was now
hoisted full-mast to welcome the guests to the proud Solli
homestead. The women, some of them beautiful in feature, were
ungraceful in form and bearing; they dressed no longer in the
picturesque Gudbrandsdal costume, but were all clothed in ill-fitting
black dresses, with no remnant of the picturesque anywhere:
queenly-looking women, some of them born, one would think, to be
mothers of Vikings; and most of them with proud pedigrees which
would excite envy in many a royal breast: shy and awkward, most of
them, even with each other. The men had perhaps a little more
savoir faire, but it was easy to see that they all led lonely lives, and
were part and parcel of that lonely land on which Nature has set a
seal of mystic melancholy. Some of the men were fine fellows, but
none as handsome as Solli, Karl, and Jens; but the Solli tribe had
long been celebrated for their good looks, and old Bedstefar in his
time had been voted the best-looking man in the whole of the
Gudbrandsdal. The guests were nearly all Bönder (landowners),

representing the best blood in the valley; most of them having the
largest Gaards, and the best-decorated pews in the churches of the
different districts. Then there was the Lensmand (bailiff), a weird old
man, rather feeble of gait, but acute in wit. He seemed much taken
with Katharine, and came several times to shake hands with her,
pretending to be a newcomer each time. But he had to keep more in
the background when his superior officer, the Foged (under-
magistrate), appeared on the scene. This gentleman was, of course,
a local personage, and he brought a very large wreath and wore an
important black satin waistcoat. There was also the doctor,
Distriktslaege[R] Larsen, famous for his rough ways and
disagreeable temper, but also for his skill in mending broken arms
and legs during the "ski" season. He seemed rather scornful of the
whole scene, but not of the port wine. And there was a Tandlaege[S]
(dentist), from Christiania, a nephew of the Sollis, who wore a very
long black frock-coat and the most fashionable pointed boots. He
was their representative man of the world and fashion, and they
prized him greatly. There were yet two other precious persons—a
member of the Storthing,[T] Bedstemor's nephew, and his wife,
rather a fine lady, who at first kept herself in 'splendid isolation,' but
soon forgot that she was a Storthingsmand's wife with a Parisian
dress, and threw her lot in with her un-Parisian-clothed relations.
She was a little suspicious of the Englishwoman, perceiving indeed a
formidable rival in well-cut garments; but directly Katharine and she
began to speak to each other in an ingenious mixture of German
and broken English, suspicions gave way to approbation, and she
said to her husband:
"Surely the English cannot be such brutes if this is a specimen of
them?"
"Pyt!" he said scornfully. "They are barbarians and brutes, all of
them."
Nevertheless he found his way over to the Englishwoman, and
was not at all eager to leave her company to join the cheerful
contingents of guests who were now strolling over to the black
house to take leave of poor Bedstefar's face. When at last he was
obliged to go, he even asked her to come too; but as Tante bravely

said, they had all seen poor dead Bedstefar often enough to satisfy
the most punctilious Gaard etiquette. Soon the Praest arrived, a
short man, with a kindly, uninspired countenance. He was
accompanied by his wife and two daughters and the Klokker (clerk),
who carried in a bag the Calvin ruff and gown still used by the
Norwegian and Danish clergymen. For it was due to the position and
dignity of the Sollis that most of the funeral service should be
conducted in the Gaard itself. If Bedstefar had been of no special
standing, he would have been taken without any preliminaries to the
churchyard, and in the absence of the clergyman, the clerk would
have said the prayers and sung the hymns, and when the clergyman
had returned from his parochial duties in some other quarter, he
would have thrown the earth and said the final words of committal
over perhaps five or six patiently waiting coffins. But Bedstefar being
who and what he was, had all possible honour shown him in his
death, as in his marriage and at his birth.
The Praest took port wine, chatted with his friends, and went with
Bedstemor to say farewell to Bedstefar. And then, at last, at last the
coffin was closed and borne through the great hall into the inner
sitting-room, preceded by the Praest, now in his vestments, and
Bedstemor, who walked bravely by his side. The nearest relations
were grouped round the coffin. The women-guests sat in the outer
room; the men stood together in the hall. The cotters, their wives,
and the servants of the house stood, some on the stairs, and some
in the porch. Tante, Katharine, and Gerda, not remembering the
custom that the men and women should be separate, sat in the hall,
and were able to see through into the inner room, where Bedstemor,
still gallantly comporting herself, joined in the dismal singing led by
the clerk, and Mor Inga, thinking of the last time that the clerk led
the singing in that very room, wept silently, and drew little Helga
closer to her side. When the singing and prayers were over, the
Praest gave a long funeral discourse, dwelling on poor Bedstefar's
virtues, which he was known not to have possessed in overflowing
measure: nevertheless tears flowed, and grim old men said, "Ja, ja,"
and the Praest was considered to have preached appropriately, and
Bedstemor seemed gratified. Then the cotters raised the coffin, bore

it out, and placed it on the low cart which had been painted black
for the occasion; and Svarten, the clever black horse who never
slipped, never failed in duty or intelligence, and knew every inch of
that winding and awkward way down to the valley, Svarten drew his
burden through the decorated gate.
"Farvel, Bedstefar," said every one.
Bedstemor stepped briskly into the carriage, together with the
Praest, Solli, and Mor Inga. The daughters remained at home to
preside over the final preparations for the feasting. The sons
followed in a cariole, and all the other men-guests helped to harness
their horses and started off leisurely in the procession, a long,
straggling, dust-raising line of about fifty conveyances. The women
stayed behind, drank coffee, and strolled about the house,
examining everything, as Ragnhild predicted; peering into the huge
old painted and decorated chests full of fine linen, looking at the old
painted sledge and cradle, dating back from 1450, precious Solli
possessions, and casting an eye on the old silver tankards, and on
the famous old carved door and sides of a pulpit, formerly belonging
to an old church which had been swept away by the falling of an
avalanche some hundred and fifty years previously. Then there were
the old painted cupboards and the queer-shaped old Norwegian
chairs and stools, and the old-fashioned, richly-carved mangles, and
the old-world slit of a recess in the wall for the Langeleik, and a fine
old Hardanger violin which Bedstefar was reported to have played
with uncommon skill; having been specially clever at giving
descriptive improvisations of Nature in her many moods, and of
things mystic, such as the song of the Huldre, and things human,
such as the ringing of marriage-bells. Alas, alas, that old-world ways
were dying out and old-world music too! Still there was much of the
old atmosphere in the Solli Gaard, and no other homestead in the
whole valley could boast of so many old-time treasures curiously
mixed up with modern importations. So that the lady funeral-guests
had much with which to amuse themselves, and they roamed into
the different bedrooms, examined Tante's possessions, and
Katharine's belongings, and did not seem at all abashed when Tante
and Katharine discovered them in the very act. Of course not, for it

was a day of entertainment; and as a sweet little old lady, a pocket
edition of Bedstemor, said, with a twinkle in her eye:
"Thou knowest we are here to enjoy ourselves. We have come a
long way. And there have not been many funerals or weddings in the
valley lately."
Knutty of course understood perfectly, and exerted herself
heroically to amuse every one, drinking coffee with every one in a
reckless fashion, and even flirting with the one man who was left
behind, an aged Gaardmand (landowner) of about ninety years. So
the time passed away cheerily for all; and when Bedstemor, Solli,
and the Praest arrived home from the churchyard, followed in due
time by the others, the feasting began. It seemed to be the
etiquette that the women should eat separately from the men. They
gathered together in the parlour, where rich soup was served to
them sitting; and after this opening ceremony, they were expected
to stroll into the great dining-room, where a huge table, beautifully
decorated with leaves, was spread with every kind of food
acceptable to the Norwegian palate: trout, cooked in various ways;
beef, mutton, veal, sauces, gravies, potatoes, even vegetables (a
great luxury in those parts), compots, and of course the usual
accompaniment of smoked mysteries. The plates, knives, and forks
were arranged in solid blocks, and the guests were supposed to wait
on themselves and take what they wished. They walked round the
table on a voyage of inspection and reflection, carrying a plate and a
fork; and having into this one plate put everything that took their
fancy, they retired to their seats, and ate steadily in a business-like
fashion. There was scarcely any talking. When the women were
served, the men came and helped themselves in the same way,
retiring with their booty either into the hall or the adjoining room. All
of them made many journeys to the generous table, returning each
time with a heaped-up plate in their hands, and in their minds a
distinct, though silent, satisfaction that the Sollis were doing the
thing in a suitable style. Every one made a splendid square meal;
but Bedstemor took the prize for appetite. She was very happy and
excited. Hers was the only voice heard. As Knutty said, it was
refreshing to know that there was at least one cheerful person

amongst those solemn one hundred and twenty guests! Knutty
herself rose to the occasion with characteristic readiness. She ate
nobly without intermission, as though she had been attending
Norwegian peasant-funerals all her life; and she gave a mischievous
wink to Gerda and Katharine every time Bedstemor rose from her
seat and strode masterfully to the table in search of further fodder.
No one offered any courtesy to any one else. It seemed to be the
custom that each person should look after herself; and there was a
look of puzzled amusement on some of the faces when Katharine
attempted to wait on one or two of the guests. Nevertheless, the
attention, once understood, was vaguely appreciated; and the pretty
little old lady whom Katharine had found in her bedroom, soon
allowed herself to be petted and spoiled by the visitors. Indeed she
abandoned all her relatives, and always sat with Knutty.
This meal came to an end about four o'clock, when there was
another relay of coffee. Some of the guests strolled about and
picked red-currants off the bushes in Bedstemor's garden. Knutty
found her way to the cowhouse and learnt from her favourite Mette
that all the servants and cotters were having a splendid meal too.
"Ja, ja," Mette said, "I have eaten enough to last for two years.
And the young ox tasted lovely! Didst thou eat of him? Ak, there is
old Kari crying her heart out because the young ox had to be killed.
Thou knowest she was fond of him. Ak, nobody has cried for
Bedstefar as much as old Kari has cried for the young ox. And she
wouldn't eat an inch of him—only think of that, Fröken, isn't it
remarkable?"
"It certainly is," said Knutty, with a twinkle in her eye. "For most
of us generally do eat up the people we love best—beginning with
the tenderest part of them."
For one moment Mette looked aghast, and then light broke in
upon her.
"Nei da," she said brightly, "but as long as we don't really eat
them, it doesn't matter, does it?"
"It is supposed not to matter," answered Knutty, moving off to
comfort old Kari, who was not only mourning for the young black ox,

but also continuing to feel personally aggrieved over her
disappointment about Clifford's ghost.
"Ak, ak, the young black ox!" cried Kari, when she saw her Danish
friend. "Eat him? Not I, dear Fröken, I was fond of him. Ak, ak!"
"Be comforted, Kari," said Knutty soothingly. "You loved him and
were good to him and didn't eat him up. What more do you want?"
"Will you tell me whether he tasted good?" asked Kari softly. "I
should like to know that he was a success."
"He was delicious," said Knutty, "and I heard the Praest and the
doctor speaking in praise of him. Of course they must know."
Kari nodded as if reassured, and disappeared into the cowhouse,
Tante's concert-room, wiping her moist eyes with her horny hands.
She came back again, and stood for a moment in the doorway.
"I cannot believe that it was not the Englishman's ghost," she
said, shaking her head mysteriously. "I felt it was a ghost. I trembled
all over, and my knees gave way."
"But you surely believe now that my Englishman is alive, don't
you, Kari?" asked Tante, who was much amused.
"I cannot be sure," replied Kari, and she disappeared again; but
Tante, knowing that she always carried on a conversation in this
weird manner, waited for her sudden return.
"That is Ragnhild's sweetheart," she said in a whisper, pointing to
a tall fair young man who had come down with another guest to
take a look at the horses. "Nei, nei, don't you tell her I told you. He
is a rich Gaardmand from the other side of the valley."
"But I have seen them together, and they don't speak a word to
each other," Knutty said.
"Why should they?" asked old Kari. "There is nothing to say."
And she disappeared finally.
"My goodness!" thought Knutty, "if all nations only spoke when
there was anything worth saying, what a gay world it would be."
Then Tante took a look at the guests' horses, some of them in the
stable, and others tethered outside, and all eating steadily of the
Sollis' corn. For the hospitality of the Gaard extended to the animals
too; and it would have been a breach of etiquette if any of the
guests had brought with them sacks of food for the horses; just as it

would have been a breach of etiquette not to have contributed to
the collection of funeral-cakes which were now being arranged on
the table in the dining-room, together with jellies, fancy creams, and
many kinds of home-made wines. Alan was sent by Mor Inga to
summon Tante to a private view of this remarkable show. Some of
the cakes had crape attached to them and bore Bedstefar's initials in
icing. They were of all imaginable shapes, and looked rich and
tempting. Tante's mouth watered.
"Ak," she cried, "if I could only eat them all at one mouthful!"
Every right-minded guest had the same desire when the room
was thrown open to the public. And all set to work stolidly to fulfil a
portion of their original impulse. Bedstemor again distinguished
herself; but Alan ran her very close. Katharine and Gerda did not do
badly. In fact, no one did badly at this most characteristic part of the
day's feasting. Then every one went up and thanked Solli and Mor
Inga, saying, "Tak for Kagen" (Thanks for the cakes). Indeed, one
had to go up and say "Tak" for everything: after wine and coffee,
dinner, dessert, and supper, which began about nine o'clock. No
sooner was one meal finished than preparations were immediately
made for the next, etiquette demanding that variety should be the
order of the day. The supper-table was decorated with fresh leaves
arranged after a fresh scheme, the centre being occupied by all the
funeral gifts of butter, some of them in picturesque shapes of
Saeters and Staburs.
Cold meats, dried meats of every kind, cold fish, dried fish,
smoked fish, and cheeses innumerable were the menu of this
evening meal. The guests did astonishing justice to it in their usual
business-like fashion; perhaps here and there Knutty remarked 'an
appetite that failed,' but, on the whole, there was no falling off from
the excellent average. Bedstemor was tired, and was persuaded to
go to bed. But she said up to the very end that she was bra', bra',
and had had a happy day. Her old face looked a little sad, and
Knutty thought that perhaps she was fretting for Bedstefar after all.
Perhaps she was.
So the first day's feasting in honour of Bedstefar came to an end.
The second day was a repetition of the first, except that the guests

began to be more cheerful. Those who lived in the actual
neighbourhood, had gone away over night and returned in the
morning; but most of them had been quartered in the Gaard itself.
Knutty talked to every one, and continued her flirtation with the
ancient Gaardmand of ninety years, who, so she learnt, had been
noted as an adept at the Halling dance. She had made him tell her
of the good old times and ancient customs, and once she succeeded
in drawing him on to speak of the Huldre. She had to use great tact
in her questionings; but, as she always said to herself, she had been
born with some tact, and had acquired a good deal more in dealing
with two generations of icebergs. So she sat amongst these reserved
Norwegians, and little by little, with wonderful patience and
perseverance, dug a hole in their frozen heart-springs. They liked
her. They said to Mor Inga:
"The fat old Danish lady is bra', bra'."
And Mor Inga whispered to her:
"Thou art a good one. They all like thee. There was a calf born
last night. We have settled to call it after thy name—Knuttyros."
"I am sure I do not deserve such an honour," Tante said, trying to
be humble.
"Yes, thou dost," Mor Inga answered with grave dignity, as she
went off to her duties as hostess.
But Tante did not understand until Clifford explained to her that a
great mark of Norwegian approval had been bestowed on her.
"Then I suppose it is like your new order of merit in England," she
said; "'honour without insult.' Ah, Clifford, I hope some day, in the
years to come, that your name will be found amongst the favoured
few."
"Not very likely, Knutty," he said. "I belong only to the rank and
file of patient workers and gropers, whose failures and mistakes
prepare the way for the triumph of brighter spirits."
"Sniksnak!" said Knutty contemptuously. "Don't pretend to me
that you are content with that. And don't talk to me about patience.
I hate the word. It is almost as bad as balance and self-control.
Balanced people, self-controlled people, patient people indeed! Get

along with them! The only suitable place for them is in a herbarium
amongst the other dried plants."
"But, Tante," said Gerda, who always took Knutty seriously, "there
would and could be no science without patience."
"And a good thing too!" replied Knutty recklessly, winking at
Katharine.
"Tante's head is turned by the unexpected honour of being
chosen as god-mother to a Norwegian cow," Clifford said. "We must
bear with her."
Knutty laughed. She was always glad when her Englishman
teased her. She watched him as he went back into the hall and sat
down near the doctor and clergyman.
"My Clifford begins to look younger again," she thought.
She watched him when Alan came and stood by him for a
moment, and then went off with Jens.
"Yes," she thought, "it is all right with my icebergs now."
She glanced across to Katharine, who was doing her best to make
friends with the women in the parlour.
"Dear one," she thought, "will you remember, I wonder, that I told
you he will never be able to speak unless you help him?"
She watched her when Alan came in his shy way and sat down
near her.
"Dear one," she thought, "the other iceberg is in love with you
too, and I am not jealous. What a wonderful old woman I am! Or is
it you who are wonderful, bringing love and happiness to us all? Ah,
that's it!"
So the second day's feasting in honour of Bedstefar came to an
end; and on the third day the men played quoits in the courtyard,
and smoked and drank more lustily. The Sorenskriver, who had had
various quiet disputes on the previous days with the doctor, the
Foged, and the Storthingsmand, now broke forth into violent
discussions with the same opponents, and was pronounced by
Knutty to be at the zenith of happiness because he was at the zenith
of disagreeableness! All the men were enjoying themselves in one
way or another; but the women sat in the big parlour looking a little

tired and bored. It was Katharine who suggested that Gerda should
sing to them.
"Sing to them their own songs," she said. "You will make them so
happy. If I could do anything to amuse them, I would. But if one
does not know the language, what can one do?"
"You have your own language, kjaere," Gerda answered, "the
language of kindness, and they have all understood it. If Tante was
not so conceited, she would know that you have really been sharing
with her the approval of the company."
"Nonsense," laughed Katharine. "Why, they think I am a barbarian
woman from a country where there are no mountains and no
Saeters! Come now, sing to them and to me. I love to hear your
voice."
"So does my Ejnar," said Gerda. "Ak, I wish he were here! He
would pretend not to care; but he would listen on the sly. Well, well,
it is good to be without him. One has one's freedom."
So she sat down and sang. She began with a little Swedish song:
"Om dagen vid mitt arbete" ("At daytime when I'm working").

[Listen]
At the day-time when I'm working,
Thou reignest in my thoughts;
At night when I am sleeping,
Thou reignest in my dreams;
At dawn when I a-wak-en,
I yearn with long-ing sore,
For my be-lov-ed sweet-heart,
So far, so far a-way.
"That is one of my Ejnar's favourites," she said, turning to
Katharine.
The company began to be mildly interested. It was not the
Norwegian habit of mind to be interested at once. Still, one or two
faces betrayed a faint sign of pleasure; and one of the men peeped
in from the hall. Then she sang another Swedish song, "Oh, hear,
thou young Dora." It was so like Gerda to feel in a Swedish mood
when she ought to have been feeling Norwegian.
The company seemed pleased. They nodded at each other.
Another man peeped in from the hall. Bedstemor strode
masterfully into the room, and sat down near the little pocket edition
of herself.
"That is another of my Ejnar's favourites," Gerda whispered,
turning to Katharine again.
She paused for a moment, thinking. No one spoke.
Then she chose a Norwegian song—Aagot's mountain-song. This
was it:

[Listen]
"O'er the hills the sun now glides,
Shadows lengthen out;
Night will soon come back again,
Folding me in her embrace;
In the stable stand the cattle,
At the Saeter door stand I.
There was a stir of pleasure in the company. Mor Inga and Solli
slipped in. Then she sang one of Kjerulf's songs, "Over de höje
Fjelde."[U]

"Fain would I know what the world may be
Over the mountains high.
Mine eyes can nought but the white snow see,
And up the steep sides the dark fir-tree,
That climbs as if yearning to know.
Ah! what if one ventured to go!"
"Up, heart, up! and away!
Over the mountains high.
For my courage is young and my soul will be gay,
If no longer bound straitly and fettered I stay,
But seeking yon summit to gain,
No more beat my wings here in vain."
The Sorenskriver came in and sat down by Katharine.
"Yes," he said, more to himself than to her, "I remember having
those thoughts when I was a young boy. What should I find over the
mountains? Ak, and what does one find in exchange for all one's
yearning?"
Gerda had sung this beautifully. The natural melancholy of her
voice suited to perfection the weird sadness of Norwegian music.
The company was gratified. They knew and loved that song well,
and some of them joined in timidly at the end of the last verse. The
old Gaardmand crept into the room and sat near Knutty.
"I could sing as finely as I could dance the Halling," he said to
Knutty, with a grim smile.
"Thou shouldst have heard me sing," said Bedstemor to Knutty. "I
had a beautiful voice."
"And so had I," said the pocket edition of Bedstemor, clutching at
Knutty's dress.
"Yes," answered Knutty sympathetically, "I can well believe it."
And she added to herself:
"We all had a voice, or think we had. It amounts to the same
when the past is past. A most convenient thing, that past—that kind
of past which only crops up when you want it!"
Then Gerda sang:

"Come haul the water, haul the wood."
This time the audience which, unbeknown to Gerda, had grown to
large proportions, joined in lustily, led by Bedstemor's cracked old
voice. She beat time, too, still playing the rôle of leading lady.
Katharine, sitting by Gerda's side, but a little in front of the piano,
saw that the hall was full of eager listeners, and that at the back of
the guests were the servants of the Gaard, including Thea and the
dramatic Mette, and some of the cotters, and old Kari. The music
which they knew and loved had gathered them all together from
courtyard, kitchen, and cowhouse. There was no listlessness on any
face now: an unwilling animation, born of real pleasure, lit up the
countenances of both men and women—an animation all the more
interesting, so Katharine thought, because of its reluctance and
shyness. It reminded her of Alan's shyness, of Clifford's too; she
remembered that Clifford had said to her several times:
"I believe I am a Norwegian in spirit if not in body; I have always
loved the North and yearned after it."
She glanced at him and caught him looking fixedly at her. He was
thinking:
"To-morrow, when she and I go off to Peer Gynt's home together,
shall I be able to speak to her as I spoke to her in my dream up at
the Saeter?"
He turned away when he met her glance, and retired at once into
himself.
Then Gerda sang other Norwegian songs, every one joining in
with increasing enjoyment and decreasing shyness: songs about
cows, pastures, Saeters, sweethearts, and Huldres, a curious
mixture of quaint, even humorous words, and melancholy music.
Finally the Sorenskriver, scarcely waiting until the voices had died
away, stood up, a commanding figure, a typical rugged Norwegian,
and started the national song:
"Yes, we cherish this our country."
Long afterwards Katharine remembered that scene and that
singing.
No voice was silent, no heart was without its thrill, no face
without its sign of pride of race and country.

CHAPTER XIX.
PEER GYNT'S STUE.
The next morning all the guests went away. They were packed in
their carioles, gigs, and carriages, and their cake-baskets were
returned to them, etiquette demanding that each guest should take
away a portion of another guest's funeral-cake offering. Ragnhild's
sweetheart was the last to go. Knutty watched with lynx eyes to see
if there was going to be any outward and visible sign of the interest
which they felt in each other; but she detected none.
"Well, they must be very much in love with each other," she said
to Gerda, "for there is not a single flaw in their cloak of sulkiness.
Ak, ak, kjaere, I am glad the funeral is over. I have not borne up as
bravely as Bedstemor; but then, of course, I have not lost a
husband. That makes a difference. Now don't look shocked. I know
quite well I ought not to have said that. All the same, Bedstemor's
strength and spirits and appetite have been something remarkable. I
believe she would like a perpetual funeral going on at the Gaard.
And how lustily she sang last evening! That reminds me, you sang
beautifully yesterday, and were most kind and gracious to the whole
company. I think Mor Inga ought to have made you the godmother
of the calf. I was proud of my Gerda. I am proud of my Gerda,
although I do tease her."
"Never mind," said Gerda, "was sich liebt, sich neckt. And I am
not jealous about the calf. I am a little jealous about the
Englishwoman sometimes. Tante loves her."
"Yes," said Tante simply, "I love her, but quite differently from the
way in which I love my botanical specimens. My botanists have their
own private herbarium in my heart."
Gerda smiled.
"I like her too, Tante," she said. "You know I was not very jealous
of her when my Ejnar began to pay her attentions."
"Because you knew they would not last," laughed Knutty. "You
need never be anxious about him. He is not a sensible human being.
He won't do anything worse than elope with a plant. Any way, he

cannot elope with Miss Frensham just now, as he is safe in the
Dovre mountains making love to the Ranunculus glacialis!"
"She told me she was going to Peer Gynt's stue with the Kemiker,"
Gerda said after a pause. "I wish I could have gone too. But my
ankle is too bad."
"Ah, what a good thing!" remarked Knutty. "That gives them a
chance. How I wish he would elope with her! But he won't, the silly
fellow. I know him. If you see him, tell him I said he was to elope
with her instantly. I am going off to the cowhouse to have a talk
with my dramatic Mette and to learn the cowhouse gossip about the
funeral-feast. So farewell for the present."
"I cannot think why Mette is such a favourite with you, Tante,"
Gerda said. "You know she isn't a respectable girl at all."
"Kjaere, don't wave the banner; for pity's sake, don't wave the
banner," Tante said. "Who is respectable, I should like to know? I am
sure I am not, and you are not. That is to say, we may be
respectable in one direction; but that does not make up the sum-
total. There, go and think that over, and be sure and keep your
ankle bad; and if you see Alan, tell him to follow me to the
cowhouse, for I want him to do something for me."
And so it came to pass that Clifford and Katharine were able to
steal off alone to Peer Gynt's stue. They had tried several times
during the funeral-feasting to escape from the company; but Mor
Inga liked to have all the guests around her, and it would have
seemed uncourteous if any of them had deliberately withdrawn
themselves. But now they were free to go where they wished
without breaking through the strict Norwegian peasant etiquette.
They had long since planned this Peer Gynt expedition. It was
Bedstemor who originally suggested it to Clifford. She was always
saying that he must go to Peer Gynt's stue; and her persistence led
him to believe that there really was some old house in the district
which local tradition claimed to be Peer Gynt's childhood's home;
where, as in Ibsen's wonderful poem, he, a wild, idle, selfish fellow
from early years upwards, lived with his mother Ãse. Clifford had not
been able to find out to his entire satisfaction whether or not this
particular stue had been known as Peer Gynt's house before the

publication of Ibsen's poem. Bedstemor had always known it as
such, and gave most minute instructions for finding it. The old
Gaardmand with whom Knutty had flirted said he had always known
it as Peer Gynt's actual home; and even old Kari, when questioned,
said, "Ja, Peer Gynt lived up over there." Bedstemor had a few
vague stories to tell about Peer Gynt, and she ended up with, "Ja, ja,
he was a wild fellow, who did wild things, and saw and heard
wonderful things."
So apparently Peer Gynt was a real person who had had his home
somewhere in this part of the great Gudbrandsdal; and Ibsen had
probably caught up some of the stories about the real man, and
woven them into the network of his hero's character. But, as Knutty
said, the only thing which really mattered was the indisputable fact
that Ibsen had placed the scene of three acts of his poem in the
Gudbrandsdal and the mountains round about, and that they—
herself, Clifford, Katharine, every one of them—were there in the
very atmosphere, mental and physical, of the great poem itself.
"And the stue stands for an idea if not for a fact," she said, "like
Hamlet's grave in my belovèd Elsinore. Go and enjoy; and forget, for
once, to be accurate."
He thought of Knutty's words as he and Katharine left the Gaard
and began to climb down the steep hillside on their way to the
valley; for Peer Gynt's home was perched on another mountain-
ridge, and they had first to descend from their own heights, gain the
valley, walk along by the glacier-river, and pass by the old brown
church before they came to the steep path which would lead them
up to their goal. He said to himself:
"Yes, Peer Gynt's stue stands for an idea in more senses than
one. Day after day, when I have not been able to open my heart to
her, I have thought that perhaps I should be able to break through
my silence on our pilgrimage to Peer Gynt's stue."
The morning was fair and fresh; summer was passing; there was
a touch of crispness in the air which suggested frost and 'iron
nights,' dreaded by the peasants before the harvest should have
been gathered in. Katharine and Clifford kept to the course of the
stream, which was a quick, though a steep, way down to the saw-

mill, beautifully situated near a foss of the glacier-river, the roar and
rush of which they heard up at the Solli Gaard. There was a bridge
across this river, and they stood there watching the tumbling mass of
water, and recalling the morning when they had passed over to the
other side on their way to the Saeters. The little Landhandleri across
the bridge was being besieged by no less than four customers. Their
carioles were fastened to a long rail outside the queer little shop
which contained everything mortal man could want, from rough
butter-boxes and long china pipes to dried cod and overalls.
"I never see these places without thinking of the isolated shops
dumped down in lonely districts out in the west of America,"
Katharine said. "Some of them were kept by Norwegians too."
"They have had their training in isolation here, you see," Clifford
said, "and so go out knowing how to cater for isolated people. And
they make a small fortune quickly and return. At least some of them
return, those in whom the love of country outweighs everything else
in life."
"I should be one of those," Katharine said. "I should always yearn
to return."
"I remember your saying you would like to bring all the broken-
hearted exiles home," he said.
"Yes," she said, "I would."
"You have a heart of pity," he said, turning to her.
"I am sorry for those who have lost their country," she said. "I
have seen them suffer. If I were a millionaire, I would find out some
of the worst cases, and give them back their country and the means
to enjoy it, or the opportunity of dying in it."
So they talked or were silent as the mood seized them. They were
happy, and frankly glad to be together alone. They left the bridge,
passed along the main road, through fragrant fir-woods, and came
to a most picturesque spot where two rivers, one of them the
glacier-river, met and rushed on together as one. They crossed this
long bridge, and found themselves on the other side of the main
valley. Here they looked back and could discern the big Solli Gaard,
perched proudly on the opposite mountain-ridge. Then their way lay
along the easy road by the winding river. It retreated from them,

Welcome to our website – the perfect destination for book lovers and
knowledge seekers. We believe that every book holds a new world,
offering opportunities for learning, discovery, and personal growth.
That’s why we are dedicated to bringing you a diverse collection of
books, ranging from classic literature and specialized publications to
self-development guides and children's books.
More than just a book-buying platform, we strive to be a bridge
connecting you with timeless cultural and intellectual values. With an
elegant, user-friendly interface and a smart search system, you can
quickly find the books that best suit your interests. Additionally,
our special promotions and home delivery services help you save time
and fully enjoy the joy of reading.
Join us on a journey of knowledge exploration, passion nurturing, and
personal growth every day!
ebookbell.com

Morphosyntactic Expression In Functional Grammar Casper De Groot

More Related Content

Similar to Morphosyntactic Expression In Functional Grammar Casper De Groot (20)

Recently uploaded (20)

Morphosyntactic Expression In Functional Grammar Casper De Groot