Polynesian Syntax and its Interfaces

Lauren Clemens (Editor)

Visit to download the full and correct content document:
https://ebookmass.com/product/polynesian-syntax-and-its-interfaces-lauren-clemens-
editor/
 OUP CORRECTED PROOF – FINAL, 13/6/2021, SPi

Polynesian Syntax and its Interfaces

OUP CORRECTED PROOF – FINAL, 13/6/2021, SPi
 OUP CORRECTED PROOF – FINAL, 13/6/2021, SPi

Polynesian Syntax and its

Interfaces
Edited by
LAUREN CLEMENS AND DIANE MASSAM

1
OUP CORRECTED PROOF – FINAL, 13/6/2021, SPi

3
Great Clarendon Street, Oxford, OX2 6DP,
United Kingdom
Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide. Oxford is a registered trade mark of
Oxford University Press in the UK and in certain other countries
© editorial matter and organization Lauren Clemens and Diane Massam 2021
© the chapters their several authors 2021
The moral rights of the authors have been asserted
First Edition published in 2021
Impression: 1
All rights reserved. No part of this publication may be reproduced, stored in
a retrieval system, or transmitted, in any form or by any means, without the
prior permission in writing of Oxford University Press, or as expressly permitted
by law, by licence or under terms agreed with the appropriate reprographics
rights organization. Enquiries concerning reproduction outside the scope of the
above should be sent to the Rights Department, Oxford University Press, at the
address above
You must not circulate this work in any other form
and you must impose this same condition on any acquirer
Published in the United States of America by Oxford University Press
198 Madison Avenue, New York, NY 10016, United States of America
British Library Cataloguing in Publication Data
Data available
Library of Congress Control Number: 2021931728
ISBN 978–0–19–886083–9
DOI: 10.1093/oso/9780198860839.001.0001
Printed and bound in Great Britain by
Clays Ltd, Elcograf S.p.A.
Links to third party websites are provided by Oxford in good faith and
for information only. Oxford disclaims any responsibility for the materials
contained in any third party website referenced in this work.
 OUP CORRECTED PROOF – FINAL, 13/6/2021, SPi

Contents

Foreword vii
List of abbreviations ix
The contributors xiii
1. Polynesian languages and their contributions to theoretical
linguistics 1
Lauren Clemens and Diane Massam
2. Gradability and modality: A case study from Samoan 11
Vera Hohaus
3. Mapping meaning to argument structure: The case of
Samoan case 36
James N. Collins
4. Deriving VOS from VSO in Tongan 61
Maria Polinsky and Eric Potsdam
5. Syntactic ergativity as absolutive movement in Tongic Polynesian 89
Lauren Clemens and Rebecca Tollan
6. Causative morphology as Voice-driven allomorphy:
The case of Samoan faʻa-causatives 113
Jens Hopperdietzel
7. Reaffirming Māori negatives as verbs 147
Sandra Chung
8. Hawaiian ai at the syntax-phonology interface 168
David J. Medeiros
9. Apparent raising in Tongan and its implications for multiple
case valuation 192
Yuko Otsuka
10. Preverbal subjects and preverbal particles: Components of the
left periphery in Māori 216
Elizabeth Pearce
OUP CORRECTED PROOF – FINAL, 13/6/2021, SPi

vi 

11. Predicate-EPP in Niuean, Tongan, and beyond 253

Julianne Doner
12. The lingering DP in Niuean 282
Diane Massam

References 303
Index 331
 OUP CORRECTED PROOF – FINAL, 13/6/2021, SPi

Foreword

In June 2018 the Department of Linguistics at the University of Toronto organized

a workshop on the syntax of Polynesian languages in honor of Diane Massam, on
the occasion of her retirement (funded by the Department of Linguistics and the
Canadian Linguistics Association). This volume was seeded at that workshop and
while its relationship to the workshop ends there, we are grateful to OUP for
allowing us (the workshop organizers) the privilege of these brief remarks com-
memorating this connection.
Like this volume, the workshop was a first of its kind, solely devoted to
Polynesian syntax. Having the first workshop on Polynesian formal syntax as a
tribute to Diane was an evocative juxtaposition as it highlighted the enormity of
her contribution to this area of study. This was evident in multiple ways. There
was, of course, the great intellectual impact of her work, which could be traced in
many of the presentations that were made. But also, and equally vital, was her
impact as a mentor and an inspiration to others. This was fully on display in the
number of students and junior scholars who came to share their work. Diane’s
own fascination with and appreciation of what one could call the genius of
Polynesian languages has always shone through her work, no doubt inspiring
others to look at these languages. Moreover, she has actively opened doors to
junior scholars, through invaluable mentorship and generosity of spirit, further
feeding the growth of research in this area.
As organizers of the workshop, and as Diane’s former mentees, we are so
excited that the seeds sown in the workshop have come to fruition in this volume,
highlighting the importance and vitality of Polynesian syntax, and in doing so,
giving us all another opportunity to reflect on Diane’s enormous contributions to
the field.

Susana Bejar
Arsalan Kahnemuyipour
OUP CORRECTED PROOF – FINAL, 13/6/2021, SPi
 OUP CORRECTED PROOF – FINAL, 13/6/2021, SPi

List of abbreviations

1 first person
2 second person
3 third person
ABS absolutive case
ACC accusative
AFLA Austronesian Formal Linguistics Association
AG agent
ANAPH anaphoric pronoun
ANTICAUS anticausative
AR apparent raising
ART article
ASP aspect
BEN benefactive
C common, case, complementizer
CAUS causative, causative prefix
CCD Constraint on Crossing Dependencies
CF corrective/contrastive focus
CL clitic
CMPR marker of comparison standard
CNTR contrastive
COM common noun
COMP complementizer
COND conditional
CONT continuity
CONTR contrary
COP copula
CSC Coordinate Structure Constraint
D determiner head
DA definitive accent
DAT dative
DEC declarative
DEF definite
DEM demonstrative
DET determiner
DIR directional particle
DIST distant (from speaker/hearer)
DIST.PST distant past
DO direct object
OUP CORRECTED PROOF – FINAL, 13/6/2021, SPi

x   

DU dual
DUP duplicate
EMPH emphatic particle
EPP Extended Projection Principle
EQ equational
ERG ergative case
ES ergative suffix
EXCL exclusive
EXIST existential
EXPL expletive
F feminine; interpretable feature
FOC focus marking
FUT future
GEN genitive
HR highest role
HUM human
IMPF imperfective
INAN inanimate
INCH inchoative
INCL inclusive
INDEF indefinite
INFN infinitive
INSTR instrumental
INTENS intensification
INTR intransitive
K case head
KP case phrase
LNK linker
L-OBJ dislocation of the object to the left
LOC locative case
LR lowest role
L-ZAS Leibniz-Zentrum Allgemeine Sprachwissenschaft
(Leibniz-Centre General Linguistics)
M masculine
m-case morphological case
NEG negative/negative predicate
NEGC negative complementizer
NFIN non-finite
NFUT non-future
NIF new information focus
NMLZ nominalization
NOM nominative case
NP noun phrase
NPI negative polarity item
NPN non-possessed noun marker
 OUP CORRECTED PROOF – FINAL, 13/6/2021, SPi

   xi

N-SPC non-specific
NSPEC non-specific
NPST non-past
OBJ object, object marker
OBL oblique
OPT optative
ORAT from the register used for the art of oratory
P preposition
PAR partitive
PART participial mood
PASS passive
PERF perfect
PERS personal article
PFV perfective
PIC Phase Impenetrability Condition
PL plural
PNI pseudo noun incorporation
POSS possessive
PP prepositional phrase
PR proper
PRED predicative/predicate marker
PREP preposition
PRN pronoun
PROG progressive
PROP proper noun
PROX proximate
PRS present
PST past
PST.PTPL past-participle
pword phonological word
Q question
QFOC focused question particle
REAL realis
REL relativizer
RP resumptive pronoun
R-SUBJ dislocation of the subject to the right
SBJV subjunctive
SDI self-directed initiator
SG singular
SPEC specific
SRC source
STAT stativizer
SUBJ subject, subject marker
SUP superlative
SVO subject verb object word order
OUP CORRECTED PROOF – FINAL, 13/6/2021, SPi

xii   

T tense
TAM tense-aspect-mood marker
TNS tense
TOP topic
TR transitive
TT theme topic
uF uninterpretable feature
USIT usitative tense
V verb
VN verbal noun
VOS verb object subject word order
VSO verb subject object word order
WCO Weak Crossover
 OUP CORRECTED PROOF – FINAL, 13/6/2021, SPi

The contributors

Sandra Chung is Distinguished Professor Emerita at the University of California, Santa

Cruz. Early in her career, she did syntactic research on Indonesian and several Polynesian
languages. The main focus of her research since 1977 has been Chamorro, an Austronesian
language of the Mariana Islands. She is the author of Case Marking and Grammatical
Relations in Polynesian (1978), The Design of Agreement: Evidence from Chamorro (1998),
Restriction and Saturation (with William A. Ladusaw, 2003), and Chamorro Grammar
(2020).
Lauren Clemens (PhD Harvard 2014) is an Assistant Professor at the University at Albany,
State University of New York, in the Department of Anthropology’s Program in Linguistics
and Cognitive Science. Her work focuses on formal syntax, prosody, and the interface
between them and draws primarily from data from the Polynesian and Mayan language
families. Her work appears in venues such as Linguistic Inquiry, Language, Natural
Language and Linguistic Theory, and Syntax.
James N. Collins is currently studying computer science at the University of Sydney. He is a
linguist interested in the representation of the syntactic and semantic components of
grammar from a formal perspective. Much of his previous work has drawn on empirical
phenomena from the Austronesian language family. He is the author of a number of papers
on formal syntax and semantics in venues such as the journal Natural Language and
Linguistic Theory, and the proceedings of Semantics and Linguistic Theory, and the West
Coast Conference on Formal Linguistics. He is currently working on an expansive inves-
tigation of phenomena relating to the interpretation of negation, especially as it interacts
with modal and gradable expressions, including neg-raising and the negation of presuppo-
sitional expressions.
Julianne Doner recently completed her PhD at the University of Toronto. In her disserta-
tion, she developed a cross-linguistic typology of Extended Projection Principle (EPP)
effects. She argued that the EPP could vary across three dimensions: argument vs. predicate
EPP, head vs. phrase EPP, or pied-piping vs. non-pied-piping EPP. She also argued that the
EPP is motivated by an anchoring requirement for language. Now, she is beginning a new
project investigating the nature of parameters through the typology of verb-initial lan-
guages, beginning with fieldwork on K’iche’ (Mayan, Guatemala). She is currently working
as a researcher and copy-editor at the University of Toronto and as a Sessional Lecturer at
the University of Guelph-Humber.

Vera Hohaus is a permanent lecturer in linguistics at the University of Manchester and an

associated researcher at the Tübingen Collaborative Research Center 833 ‘The Construction
of Meaning’. She received her training at the Eberhard Karls Universität Tübingen (from
which she was awarded her PhD in 2015) as well as the University of Massachusetts,
OUP CORRECTED PROOF – FINAL, 13/6/2021, SPi

xiv  

Amherst, and the University of British Columbia, Vancouver. Her research specializes in
semantics and cross-linguistic variation. She has been doing fieldwork on the grammar of
Samoan (Austronesian, Oceanic) since 2009.
Jens Hopperdietzel is a post-doctoral researcher at the University of Manchester. Prior to
this position, he was affiliated with the Leibniz-Centre General Linguistics (L-ZAS) and the
Humboldt University of Berlin, where he recently completed his dissertation. In his
dissertation, he investigates the event and argument structure of resultative serial verb
constructions in Oceanic languages—with a focus on the Polynesian language Samoan and
the Vanuatu language Daakaka—and its implications for the syntactic and semantic
composition of resultatives cross-linguistically. More generally, his research centers around
syntax and its interfaces to morphology, semantics, and prosody, with a strong commit-
ment to comparative research and fieldwork.
Diane Massam (PhD MIT, 1985) is Professor Emeritus in the Department of Linguistics at
the University of Toronto. Her main research interests are argument structure, case,
predication, word order, and nominal structure, with a focus on the Niue language and
on register in English. She has published many articles on these topics as well as a recent
book Niuean: Predicates and Arguments in an Isolating Language (OUP, 2020). She is also
the editor of Count and Mass Across Languages (OUP, 2012) and, with Jessica Coon and
Lisa deMena Travis, of The Oxford Handbook of Ergativity (OUP, 2017).

David J. Medeiros is an associate professor in the Department of Linguistics/TESL at

California State University, Northridge. In addition to research on Hawaiian and
Polynesian linguistics, recent and forthcoming papers include ‘Imperatives Under
Coordination’ (with Ezra Keshet) in Natural Language and Linguistic Theory and ‘Ceiling
Effects on Weight in Heavy NP Shift’ (with Paul Mains and Kevin B. McGowan), to appear
in Linguistic Inquiry.
Yuko Otsuka is a professor of linguistics at Sophia University, Tokyo. Her research
interests are generative syntax, and Polynesian and other Austronesian languages. She
has worked extensively on Tongan with a special focus on case and ergativity. She is also
a co-editor of Oceanic Linguistics (University of Hawaiʻi Press).
Elizabeth Pearce gained her PhD at the University of Illinois in 1985. She is an honorary
senior fellow at the University of Melbourne and was formerly a senior lecturer in
linguistics at Victoria University of Wellington. Her primary research interests are in
syntax and historical linguistics. She has worked on aspects of the syntax of Romance
languages and of Oceanic languages, and she has carried out fieldwork in Vanuatu. She is
the author of two books: Parameters in Old French Syntax: Infinitival Complements (1990)
and A Grammar of Unua (2015).
Maria Polinsky is Professor of Linguistics at the University of Maryland, College Park and
Director of the National Heritage Research Center at UCLA. Her research combines the-
oretical syntactic work with in-depth investigation of understudied languages. Her interest
in ergativity has led her to do primary work on Austronesian languages and a number of
languages spoken in the Caucasus. She has served on the editorial boards of multiple
journals, on the National Science Foundation’s Expert Panel on Linguistics, and on
 OUP CORRECTED PROOF – FINAL, 13/6/2021, SPi

  xv

advisory boards of several international centers. She is the author of over a hundred
scholarly articles and several books including Deconstructing Ergativity (2016)
and Heritage Languages and Their Speakers (2018).
Eric Potsdam is Professor of Linguistics at the University of Florida, where he has been
since 2000. He has published widely on the syntax of English and Austronesian languages,
particularly Malagasy, the native language of Madagascar, and the Polynesian languages.
Topics that he has worked on include the structure of non-declarative clauses (interroga-
tives, imperatives, and exclamatives), raising and control, and a range of potentially
elliptical constructions such as sluicing, exclamatives, comparatives, and extraposition.

Rebecca Tollan completed her PhD at the University of Toronto and is currently Assistant
Professor of Linguistics at the University of Delaware. Her research interests include the
syntax of long-distance dependencies, case, and argument structure and how these phe-
nomena are processed in real-time sentence comprehension. Much of her recent work has
examined the processing of wh-questions and resolution of anaphoric pronouns in Niuean.
OUP CORRECTED PROOF – FINAL, 13/6/2021, SPi
 1
Polynesian languages and their
contributions to theoretical linguistics
Lauren Clemens and Diane Massam

1.1 Overview

This volume presents research in theoretical syntax and its interfaces with
semantics and prosody based on a range of Polynesian languages. There are few
such volumes, as generally Polynesian languages are studied within the larger
domain of Austronesian languages (e.g. through conferences such as the
International Conference on Austronesian Linguistics, the Austronesian Formal
Linguistics Association (AFLA), the International Austronesian and Papuan
Languages and Linguistics Conference, and in the journal Oceanic Linguistics),
or within the domain of Oceanic languages (e.g. through the Conference on
Oceanic Languages and in the volume Lynch et al. 2002). In both the
Austronesian and the Oceanic families, Polynesian languages form a minority,
so the issues they raise are often overshadowed by issues relevant to a larger
proportion of Austronesian languages (e.g. historical reconstruction and the
pressing need for comprehensive description), or by issues relevant in
Austronesian languages with larger populations of speakers (e.g. voice systems
and extraction restrictions).
While Polynesian languages share many linguistic properties with the lan-
guages in these larger groupings, they also present unique characteristics and
theoretical opportunities. For example, with their relatively shallow time depth,
low-level position in the language family, and their almost complete geographical
isolation from each other, they present an ideal laboratory for the study of
parametric micro-variation. Of course, such study rests on solid theoretical
analyses of individual languages. This volume focuses on issues that are uniquely
Polynesian, and it includes chapters that provide in-depth analyses of these
issues within particular languages, as well as chapters that take on comparative
challenges within the language family. The volume includes work on theoretical
syntax, semantics, morphology, and prosody across the sub-groups of Polynesian,
with chapters focusing on Hawaiian, Māori, Niuean, Samoan, and Tongan.
OUP CORRECTED PROOF – FINAL, 13/6/2021, SPi

2     

1.2 Polynesian languages

For the most part, Polynesian languages are spoken in Polynesia, which is
generally defined as a triangle of islands in the Pacific, with Hawaiʻi,
New Zealand (Aotearoa), and Easter Island (Rapa Nui) forming the three most
distant points. According to Lynch et al. (2002), Polynesia was settled around
1000 , ultimately resulting in nineteen languages being spoken in the region. In
addition, there are fifteen Polynesian Outlier languages spoken in Melanesia and
Micronesia, by people who moved away from Polynesia at later dates after
originally settling there. The languages cover a large geographic area, but they
form a coherent group through their grammatical innovations (Lynch et al. 2002;
cf. Pawley 1966, 1967; Biggs 1971; Clark 1979). The Polynesian family lies within
the Central/Eastern branch of the Oceanic languages, and it has two main
branches, Tongic and Nuclear Polynesian, with the latter being further subdi-
vided, as in Figure 1.1, which provides a standard family tree. The main languages
of discussion in this volume, Hawaiian, Māori, Niuean, Samoan, and Tongan,
represent each branch of the tree (with the exception of Rapanui).

Polynesian

Tongic Nuclear
Polynesian
Tongan,
Niuean

Samoic/ Eastern
Outlier Polynesian
Samoan, Tokelauan,
Tuvaluan, Nukuoro, Rapanui Central
Kapingamarangi, Nukuria, Eastern Polynesian
Takuu, Nukumanu,
Luangiua, Sikaiana,
Pileni, East Uvean, Marquesic Tahitic
East Futunan, Pukapuka,
Mangarevan, Tahitian,
Rennell-Bellona, Tikopia,
Marquesan, Tuamotuan,
West Uvean, West
Hawaiian Rapa, Penrhyn,
Futuna, Ifira-Mele,
Rarotongan, Māori
Emae, Anuta
Moriori (extinct)

Figure 1.1 Polynesian language family

Source: adapted from Lynch (1998).
 OUP CORRECTED PROOF – FINAL, 13/6/2021, SPi

     3

Each of these languages has a relatively small number of speakers, and most of the
languages are considered to be endangered (http://www.unesco.org/languages-
atlas/).

1.3 Early history of Polynesian linguistics

The Western study of Polynesian languages began in essence with Cook’s

voyages (1768‒80) during which Sir Joseph Banks made observations about
the commonalities of languages in diverse areas such as New Zealand and
Tahiti (Biggs 1974). The 18th and 19th centuries saw the publication of lexicons,
grammar sketches, religious works, and bibles, often by missionary-linguists,
and many of these remain as key reference works today (e.g. Davies 1851;
Andrews 1865; cf. Sebeok 1971). Following on from this were works establishing
the historical relations among the languages (Elbert 1953; Capell 1962; Grace
1968), a pursuit which remains robust up to the present day, both for Polynesian
and the larger family of Austronesian.¹ A notable accomplishment is POLLEX
Online, a large-scale comparative dictionary of Polynesian languages, the produc-
tion of which is overseen by Bruce Biggs, Andrew Pawley, and Ross Clark (Greenhill
& Clark 2011).
Work on Polynesian languages died down in the early part of the 20th century,
but modern linguistics turned its attention to Polynesian languages in the 1950s,
as noted in Biggs (1971). Much of this work continued to focus on phonology and
morphology, with an eye to historical reconstruction, and we will not review this
impressive work here (see books and papers by the authors in note 1, and their
references). Unusual among the contributions through this time period, in its
focus on syntactic reconstruction, is Clark (1976). Clark’s main goal is recon-
struction of Proto-Polynesian, but along the way he presents a careful study of
many of the principal typological characteristics of Polynesian languages, as well
as several key constructions which remain of central interest to theoretical lin-
guists today. The most salient typological characteristics of Polynesian languages²
are their isolating, particle-based morphology, sentence-initial Tense/Aspect/
Mood (TAM) marking and negation marking, variation in word orders (mainly
VSO/VOS/SVO), varying case systems (accusative and ergative), cleft structures,

¹ Of note is the work of K. Alexander Adelaar, Juliette Blevins, Robert Blust, Ross Clark, Terry
Crowley, Otto Dempwolff, Isidore Dyen, Paul Geraghty, George Grace, Roger Green, Ray Harlow, John
Lynch, Jeff Marck, G. B. Milner, Andrew Pawley, Malcolm Ross, Mary Walworth, William Wilson,
John Wolff, David Zorc, among others. See Otsuka (2005e, 2006a) for a specifically Polynesian
discussion of the history of the language family.
² We focus on syntax, but recent works within theoretical Polynesian phonology include:
Taumoefolau (2002), Anderson & Otsuka (2006), Rolle & Starks (2014), Zuraw (2018), and Zuraw
et al. (2014); and on phonology and its interfaces: de Lacy (2004); Calhoun (2015, 2017); and Yu &
Stabler (2017).
OUP CORRECTED PROOF – FINAL, 13/6/2021, SPi

4     

reduplication, and complex genitive constructions. The study of these properties

has led to theoretical insights in topics such as the morphological, syntactic, and
prosodic nature of lexical vs functional words, configurationality and move-
ment, the architecture of predicate-argument relations, the structure of the left
periphery, the role of case marking, and the syntax of noun phrases. These areas
and related topics currently form the major focus of work in Polynesian syntax,
as we see reflected in the chapters in this volume, which are discussed in
Section 1.5.
Of course, the major contribution to the understanding of Polynesian lan-
guages, both historically and in the present day, has been made by speakers of
the various Polynesian languages. As consultants and teachers, they have provided
invaluable insights both to linguists and to their fellow-speakers; and as linguists,
they have contributed substantially to the literature, through dictionaries, gram-
mars, and dissertations (only a few of which are cited here), as well as within the
field of theoretical linguistics. Not least, many of these language professionals are
also working on the front lines of language maintenance and revitalization.

1.4 Contributions to modern theoretical linguistics

Polynesian languages became important early on within theoretical linguistics,

largely due to debates about the nature of case and ergativity, passivization, and
subjecthood. These came to the fore through the work of Hale (1968) and Hohepa
(1969a), as well as Pawley (1973); Biggs (1974); Chung (1974, 1977, 1978);
Sharples (1976); Sinclair (1976); Wilson (1976); Cook (1978); Hawkins (1979);
and Reedy (1979). These works, many of which were published in The Journal
of the Polynesian Society, had the effect of bringing Polynesian grammar into
mainstream discussions (and homework assignments) within Government and
Binding, Relational Grammar, and Lexical Functional Grammar (as well as into
functional and descriptive discussions), at a time when generative linguistics was
first expanding to include lesser studied languages (e.g. Chung & Seiter 1980; Cole
et al. 1980; Seiter 1980; Hooper 1984; Levin & Massam 1985, 1988; Massam 1985;
Waite 1987; Cook 1988, 1991; and Lillo-Martin 1989). Throughout the same time
period, Biggs (1974) and Krupa (1982), in their discussion of Polynesian mor-
phology, brought forward questions about the nature of words, prosody, and word
classes in Polynesian, also a major topic for current theoreticians, while Ochs
(1988) raised questions about language acquisition.
More recently, particularly within Minimalism, Polynesian languages have
become important for the study of other theoretical issues, such as pro drop,
V-initial and other word orders, complex predicates, inversion, predication,
ergativity, noun incorporation, extraction and resumption, reflexives, modality,
question formation, and the structure of the left periphery in both nominal and
 OUP CORRECTED PROOF – FINAL, 13/6/2021, SPi

     5

sentential domains (e.g. Lazard & Peltzer 1991; Dukes 1996; Hawkins 2000;
Hooper 2000; Bauer 2004; Custis 2004; Hendrick 2005a; Ball 2008; Koopman
2012; Macdonald 2014; Brown & Koch 2016; Kikusawa 2017; Douglas 2018).³
This theoretical work has been made possible in part by the important reference
grammars that have brought the features of the languages forward, including
grammars on Fakauvea or Wallisian (Moyse-Faurie 2016), Hawaiian (Elbert &
Pukui 1979), Māori (Ngata 1901 and later editions; Bauer 1993, 1997; Harlow
2007), Pukapukan (Salisbury 2002), Rapanui (du Feu 1996), Samoan (Mosel &
Hovhaugen 1992), Tongan (Churchward (1953), Tuvaluan (Besnier 2000), Ùa
Pou, a Marquesan dialect (Mutu with Teìkitutoua 2002), and Vaeakau-Taumako
or Pileni (Næss & Hovdhaugen 2011), as well as many others. In addition to these
grammars, there is also a wealth of excellent bilingual and unilingual dictionaries,
many developed by speakers or teams of speakers of Polynesian languages, which
are too numerous to outline here. Polynesian languages have also been important
in typological linguistics (e.g. papers in Fischer & Sperlich 1999; Fischer (2000);
and Bril & Ozanne-Rivierre 2004).
The blossoming of theoretical work on Polynesian languages was fostered by
the foundation in 1994 of AFLA. This conference has continued to provide an
annual venue for sharing formal research and for collaboration amongst scholars
such as those represented in this volume, creating a rich background for new
generations of theoretical Polynesianists. Polynesian linguistics continues to grow
and to impact our understanding of phonology and syntax, and, more recently, of
prosody and semantics as well. This volume comes at a vibrant time in the field of
theoretical Polynesian, and it represents the work of both seasoned and new
scholars in the area. In Section 1.5, we overview the chapters included in this
volume, and show how each chapter interacts with the issues discussed in
Sections 1.3 and 1.4.

1.5 Contents of the volume

The chapters in this volume address topics ranging in scope from the clausal
architecture of Polynesian languages to fine-grained analyses of single lexical
items in individual languages. In all the chapters, authors focus on issues that
are distinctive to Polynesian languages and relevant across the entire family; and,
in all cases, they connect the themes to broader questions in linguistic theory. Here
we present an outline of each chapter, briefly noting its significance within
Polynesian and theoretical linguistics.

³ This list is by no means exhaustive, for example, we do not cite here the foundational and ongoing
work by the participants in this volume, as much of it is reviewed or cited in the chapters themselves.
See the References section in this volume for a fuller bibliography.
 OUP CORRECTED PROOF – FINAL, 13/6/2021, SPi

6     

In Chapter 2, Hohaus presents a unified analysis of Samoan sili, a predicate

used to express both weak necessity and the superlative. In this groundbreaking
study of Polynesian modality, she argues for a gradable, degree-based relational
lexical entry for sili (rather than one based on quantification over possible worlds),
which allows for both its modal and non-modal meanings. She shows how the
Samoan morpheme can teach us about the relationship between gradability and
modality, as well as about the different compositional pathways that languages can
take to construct weak necessity, priority modal meanings, as well as superlative
meanings, which, in Samoan, build on the meaning of the Positive form. Her
study initiates discussion of Polynesian modality, explores context dependency
(cf. Hopperdietzel, Chapter 6, this volume), and illustrates how cross-linguistic
analysis can extend (and test) our understanding of core semantic concepts.
In Chapter 3, Collins also examines an issue at the crossroads of syntax and
semantics in Samoan, developing a general theory of how morphological case
relates to lexical semantics. A tantalizing question in general, this takes on
particular salience in the ergative Polynesian languages where there is almost,
but not quite, a one-to-one mapping between ergative case and agentivity. Samoan,
like many Polynesian languages, exhibits more than one morpho-syntactic encod-
ing for transitive predicates, known as transitive and middle constructions, with the
case patterns - and -,⁴ raising questions about how verbal meanings
are paired with argument structures. Collins develops an analysis of case that
depends on verbal semantics, arguing that to appear in the - frame, first,
a verb must have a self-directed initiator argument; and, second, the less agentive
argument cannot be a goal. His chapter contributes to discussions of case in
Polynesian (cf. Clemens & Tollan, Chapter 5, and Otsuka, Chapter 9, both this
volume) and he proposes an analysis within Optimality Theory, thus contributing
to studies of how OT can apply to syntactic structures.
From the mapping between thematic arguments and cases, we turn to the
surface syntactic positions of the arguments themselves. In Chapter 4, Polinsky &
Potsdam present a novel account of word order variations in Tongan. They adopt
Otsuka’s (2000) V⁰-raising account of Tongan verb-initiality, but argue against the
widely accepted A-scrambling treatment of VOS (Custis 2004; Otsuka 2005b; see
also Clemens & Tollan, Chapter 5, and Otsuka, Chapter 9, both this volume).
Instead, Polinsky & Potsdam develop a number of arguments in favor of a
rightward subject displacement account of Tongan VOS. Their arguments for
this proposal are drawn from discourse interpretations as well as from syntactic
considerations such as reflexive connectivity and word order. They consider a
subject-final base-generated order, but argue in favor of movement. They con-
clude with general discussion of the role of rightward movement in grammar,

⁴ The case Collins calls  is variously referred to in the literature as middle case (e.g. Chung 1978),
as accusative case (e.g. Tollan 2018), or (in Niuean) as locative or goal case (Massam 2020).
 OUP CORRECTED PROOF – FINAL, 13/6/2021, SPi

     7

noting that while some linguists rule it out altogether, it continues to have an
analytic tradition of its own. Their chapter contributes to the ongoing discussion
about VSO/VOS word order variations in Polynesian, while also addressing larger
issues about movement typology in linguistic theory.
Also contributing to the ongoing discussion about VSO/VOS word order
alternations in Tongic languages is Chapter 5 by Clemens & Tollan. Building on
literature connecting the occurrence of syntactic ergativity to the locus of absolu-
tive case assignment (e.g. Coon et al. 2014), Clemens & Tollan develop an account
of the alternating word order in Tongan as compared to the strict VSO order of
Niuean. They argue that in Tongan, the absolutive argument must A-move
leftwards for absolutive case (which they argue is high, on T in Tongan), and
subsequently it can be pronounced in either its base position or its case position,
accounting for the word order alternation. Assuming this analysis, the presence of
syntactic ergativity (i.e. the restriction of movement to only the absolutive argu-
ment) falls out in Tongan, since the ergative argument is ‘trapped’ in the lower
position by the fronted absolutive, due to constraints on crossing dependencies. On
the other hand, they argue that absolutive case is assigned low, by voice in Niuean,
so there is no movement of the absolutive argument, and the ergative argument is
thus free to extract, resulting in the lack of syntactic ergativity in Niuean. Their
chapter thus unifies several micro-parametric variations between Tongan and
Niuean; and, further, it builds bridges with related work on case and syntactic
ergativity in these languages as well as others, such as those in the Mayan family.
A number of chapters in this volume address the functional structure of the
clausal spine. Within the verbal domain, Hopperdietzel’s Chapter 6 illustrates a
bundling paradox with the Samoan causative morpheme faʻa (a morpheme
appearing across Polynesian languages), which shows both voice-bundling and
non-bundling properties. He argues that this paradox reveals the locus of causa-
tive semantics in the derivation, arguing for a contextual approach (cf. Hohaus,
Chapter 2, this volume), rather than the designation of a causative head. He
proposes that what might appear to be an instance of morphological bundling
in Samoan is instead a case of bi-directional contextual allomorphy, where the
eventive v head is spelled out as faʻa but only in the presence of a higher voice
head. He argues that this explains why causatives, but not inchoatives, are
morphologically marked. Not only does Hopperdietzel present an in-depth
study of faʻa, which is one of the key Polynesian prefixes (in contrast to the
many post-verbal particles), he also discusses the causative alternation in Samoan
in the context of apparent bundling phenomena cross-linguistically (Pylkkänen
2008; Harley 2017).
Chapter 7 by Chung adjudicates between competing analyses of Māori nega-
tion: one in which it is a verb and one in which it is a functional head. Returning to
earlier work by Hohepa (1969b); Chung (1970); and Bauer (1997), she argues that
Māori negation is a lexical head, either V or A, and it serves as the main
OUP CORRECTED PROOF – FINAL, 13/6/2021, SPi

8     

intransitive predicate of a clause. Further, she explains why Māori negation must
be verbal, in spite of the (perhaps false) assumption that in most languages NEG is
a functional head. Her account relies on a number of interconnected observations
from the interfaces based on previous work on Māori prosody, including the fact
that negation precedes the predicate in verb-initial languages; that lexical heads
correspond to prosodic words whereas functional heads are deficient; and the
requirement that TAMs must combine with a minimal prosodic word to their
immediate right (cf. Biggs 1961; Bauer 1993; de Lacy 2003; Herd 2003; and Brown
2015). Chung’s chapter thus contributes to ongoing discussions about wordhood
in Polynesian; to cross-linguistic studies of negation and clausal architecture;
and to the body of current work that integrates the domains of semantics, syntax,
and prosody, a rich area of promise in Polynesian studies.
The next two chapters in this volume discuss cross-clausal dependencies. In
Chapter 8, Medeiros, like Chung, takes a syntax-phonology interface approach,
examining the morpheme ai in Hawaiian, which also appears in many Polynesian
languages (Chapin 1974). The morpheme ai is often considered to be a resumptive
pronoun, as it appears in or near the gap position of certain moved constituents,
but Medeiros argues that it is not a pronoun, having no lexical features (such as
phi-features). Instead, its occurrence is accounted for entirely by syntax, which
explains why it is never optional, and only appears in a sub-set of constructions.
Medeiros considers that ai occurs when any non-subject, or any non-local subject,
is displaced as a repair for an illicit linearization output. The insertion of ai is thus
a last-resort mechanism to rescue conflicting instructions at the syntax-phonology
interface, similar to some analyses of ellipsis (Fox & Pesetsky 2005). He argues that
similarities between ai and resumptive pronouns in other languages is due to
general properties of locality, with both involving cyclic spell-out domains.
Medeiros’s linearization algorithm accounts for ai, as well as for main clause
word order and VP-remnant formation. The chapter contributes to current
models of the syntax-phonology interface, and it also lays the groundwork for
further study of the morpheme ai, which presents interesting micro-parametric
variation across Polynesian languages.
Turning from apparent resumption to apparent raising (AR), in Chapter 9,
Otsuka tackles the conflicting A- and A-bar-like properties of embedded DPs in
Tongan that move to a position where they receive matrix case, but also exhibit
long-distance dependencies and sensitivity to islands. These have been analyzed as
cases of movement of a null operator, but Otsuka argues instead for an analysis
where the overt argument bears a topic feature and moves to an embedded CP
specifier, where it undergoes case valuation from a matrix v, made possible
through a multiple case valuation mechanism. She argues that her analysis can
account for the properties of AR such as the impossibility of pronoun movement
(due to cliticization requirements) and of movement of an ergative (via syntactic
ergativity), as well as the A-bar properties of the movement. Otsuka presents
 OUP CORRECTED PROOF – FINAL, 13/6/2021, SPi

     9

a detailed analysis of multiple case marking in which a DP can remain active even
with a valued case feature if it is located at the edge of a phase, contributing to
theories which separate abstract case as a licensing mechanism from the morpho-
logical case. Otsuka’s chapter continues the long-standing discussions about raising
constructions in Polynesian (going back to Seiter 1980 and Massam 1985), which
test the validity of standard theoretical assumptions about movement, the clear
separation of A- and A-bar operations, and the nature of case-marking.
Pearce’s Chapter 10 addresses the spinal architecture of Polynesian languages
through a study of preverbal subject constructions in Māori and their interactions
with the TAM system, arguing that preverbal subjects must be preceded by a
T-marked item, attributed to a general T-initial (vs V-initial) requirement in
Māori. In her examination of topic and focus constructions she develops a
cartographic analysis of the CP domain of Māori (cf. Rizzi 1997; Benincà &
Poletto 2004). While Polynesian focus constructions are widely considered to be
clefts, they are problematic in containing neither an expletive nor a copular verb.
Pearce develops an analysis in which there are tense and predicate projections
within the left periphery, bolstering the traditional cleft analysis. She also supports
the proposal that negation in Māori is verbal (Chung, Chapter 7, this volume) and
the view that Tense and Case are intertwined (Massam 2000; Doner, Chapter 11,
this volume; cf. also the hybrid nature of Case and Det in Niuean nominals
discussed in Massam, Chapter 12, this volume). Her chapter explores the key
issues of word order, cleft constructions, and TAM systems in Polynesian, and it
also augments the cross-linguistic study of the cartography of the left periphery.
One of the foremost ways work on Polynesian languages has left its mark on
syntactic theory is in the idea that languages differ with respect to how EPP
(Extended Projection Principle) on T is satisfied (Massam & Smallwood 1997).
Doner’s Chapter 11 focuses on Niuean, and builds on previous proposals for
variation in EPP features (Alexiadou & Anagnostopoulou 1998; Davies &
Dubinsky 2001; Biberauer 2010) to develop a typology in which the EPP is
alternatively satisfied by DP, D features, or the predicate. She argues for a
predicative EPP, in contrast to a verbal one, on the grounds that non-verbal
predicates can satisfy the EPP, but light verbs cannot, in conjunction with the
fact that in languages that have it, verb-raising does not satisfy the EPP. Making
reference also to Irish, French, and Inuktitut, Doner presents a clustering of
properties that are common to predicative EPP languages, all related to the
existence of a high impoverished T head, where T is merged in C (cf. Massam
2000; Pearce, Chapter 10, this volume), often echoed by the presence of a defective
D as well (Massam, Chapter 12, this volume). She extends her analysis to Tongan
(cf. Otsuka 2005a). In her study, she ties the properties of the Polynesian left
periphery to the V-initial word order of these languages (cf. Chung, Chapter 7,
and Pearce, Chapter 10, both this volume), as well as presenting a typology of the
EPP, and raising questions about its nature.
OUP CORRECTED PROOF – FINAL, 13/6/2021, SPi

10     

The final chapter in the volume, Chapter 12 by Massam, also looks at the left
periphery, but here it is the DP that is the object of study, with a focus on Niuean
as compared with Tongan. Niuean pre-nominal markers of case are cognate with
Tongan articles, leading to confusion about their status as case markers or articles.
Massam compares the nominal particle systems of Niuean and Tongan, and
proposes that the Proto-Tongic case markers disappeared in Niuean, and were
replaced by forms that are cognate with the Proto-Tongic articles, synchronically
merged in KP (case phrase). The result is the loss of a systematic determiner
system in Niuean, but Massam argues that this loss did not result in the loss of
D (and DP) in Niuean. Instead, D remains as the locus of the feature [+/ proper],
with which the higher case markers agree (cf. C and T discussed in Pearce,
Chapter 10, and Doner, Chapter 11, both this volume). As further evidence, she
argues that D can host phrases in its specifier, which contribute meanings associated
with (in)definiteness and quantification, and further, that D houses a linker when its
specifier is filled. Finally, she posits that there are two occasional articles (negative
polarity ha and personal a) that merge in D. Her chapter shows how small
historical changes can lead to reanalysis of entire systems, and it clarifies a puzzle
in comparative Polynesian about the status of Niuean pre-nominal particles. The
chapter also contributes to theoretical debates about the universal nature of DP
and its status in languages that do not exhibit determiner systems.

1.6 Concluding remarks

This concludes our overview of the contributions of Polynesian languages to

linguistic theory and our introduction to the chapters that follow. We would like
to extend our deepest appreciation to Susana Bejar, Arsalan Kahnemuyipour, Lisa
Matthewson, Yves Roberge, Lynsey Talagi, and Heidi Quinn. We would also like to
thank the editorial team at Oxford University Press, especially Vicki Sunter for her
guidance throughout, and Joy Mellor, Saraswathi Rajan, and Lydia Shinoj, and the
anonymous reviewers who reviewed the chapters. We consider that the chapters
herein illustrate the richness of the contribution that the study of Polynesian
languages makes to the larger world of comparative and theoretical linguistics.
 2
Gradability and modality
A case study from Samoan
Vera Hohaus

2.1 Introduction

Possibility and necessity are gradable notions: There might be only an ever so
slight possibility that global warming will stay below the 2°C mark, whereas there
is a good possibility that this will not happen. We should, or rather, must reduce
our carbon footprint significantly. While the discussion of gradable modality
already features in Moore (1900) and Russell (1903), and, more prominently, in
Lewis (1970) and Kratzer (1981), it has only recently been the focus of a more
sustained research effort in linguistics (Villalta 2008; Yalcin 2010; Lassiter 2011,
2017; Rubinstein 2012, 2014; Pasternak 2016; Portner & Rubinstein 2016), which
centers around the following two questions: First, are all modal expressions in
English gradable? Second, what is the linguistic representation of this gradability,
and how does it compositionally interact with other linguistic material? As for
English, the very least we can say is that some of its modal expressions may occur
in comparison constructions, while others may not, as shown in (1) and (2).

(1) a. I need to go on vacation more than I need to finish this work.

b. It is very likely that we have missed our train.
c. are completely unable to fulfill your request.
d. It is 95 percent certain that our team will win. (Solt 2015: 20, no.(18))

(2) a. ?You must call Zoe more than you must call Barbara.
(Portner & Rubinstein 2016: 261, no.(11a))
b. ?It is more possible now. (Klecha 2014: 2013, no.(2a))

Should at least the examples in (1) (or potentially all, despite differences in
acceptability) thus receive a degree-based analysis as is standardly adopted for
gradable adjectives and comparison constructions in English (von Stechow 1984;
Kennedy 1997; Heim 2001; Beck 2011)? Under such a degree-based approach,
gradable adjectives like tall are commonly analyzed as relational, as in (3), with a
OUP CORRECTED PROOF – FINAL, 16/6/2021, SPi

12  

range of functional heads (like the comparative) operating on the degree argument
which the adjective makes available. Such an analysis does not, however, line up
straightforwardly with the quantificational treatment of modality along the lines
of (4), under which necessity is analyzed as universal and possibility as existential
quantification over possible worlds (Kratzer 1981, 1991, 2012), more specifically
over the best of the accessible worlds in the context relative to at least one ordering
source.
pffiffiffiffiffiffiffi
(3) ⟦ tall⟧ ¼ λw:λd:λx:(w)(x) ≥d

(4) ⟦must⟧ ¼ λw:λp〈s;t〉 :∀w′ ∈ order(w)( ∩ access(w)) : p(w′) = 1

⟦can⟧ ¼ λw:λp〈s;t〉 :∃w′ ∈ order(w)( ∩ access(w)) : p(w′) = 1

Assigning a degree-based semantics to modal expressions (under which they

relate degrees and propositions) requires that we address the nature of the
underlying scale, and different authors have put forward different proposals as
to how to define such measures of necessity and possibility (Lassiter 2011; Klecha
2014; Pasternak 2016; Portner & Rubinstein 2016).
This chapter aims at making a first step towards adding a cross-linguistic
perspective to these research questions, especially as most if not all research on
gradability and modality is based on well-studied Indo-European languages,
and in particular, English and German. We contribute here a case study from
Samoan, an Austronesian verb-initial language, which is spoken on the South
Pacific islands comprising the two Samoas—the Independent State of Samoa and
American Samoan, an unincorporated territory of the United States of America.
More specifically, this chapter provides a description and analysis of the verb sili,
based on data from the author’s fieldwork since 2009.¹ This verb is both syntac-
tically as well as semantically fairly versatile: It can take proposition-type
arguments headed by the complementizer ona, as in (5a) and (6a), but also
individual-type noun phrase arguments, as in (5b) and (6b). In both of these
syntactic configurations, it may then be used to express weak necessity relating to
the modal category of priority (in the terminology of Portner 2009), as is the case
in (5). The verb is, however, also one of the preferred strategies for expressing
superlative meanings, as in (6), despite the literal translations provided (and one
of the goals of this chapter will be to address this mismatch between form and
meaning). Milner (1966: 209) suggests to translate the verb as “to be highest, to be

¹ The data presented here consist of naturally occurring as well as elicited example sentences.
Elicitation involved translation, acceptability judgment and production tasks (see also Matthewson
2004, 2011). We have preserved the original orthography of the examples, resulting in some variation
with respect to diacritics and word boundaries.
 OUP CORRECTED PROOF – FINAL, 16/6/2021, SPi

   13

top, to be topmost, to be great, to be principal, to be best, to be preferable”.

Throughout, we adopt ‘to be great’ as a translational approximation.

(5) a. E sili [ona ʻe nofo i lalo].

 SILI that you sit  down
‘You should sit down.’ (Lit. ‘It is great that you sit down.’)
b. E sili [le nofo sāuniuni nei mō taeao].
 SILI the life prepare today for tomorrow
‘Prepare for tomorrow, today!’
(Lit. ‘The life which prepares today for tomorrow is great.’)

(6) a. E sili [ona umi Sina].

 SILI that tall Sina
‘Sina is the tallest.’ (Lit. ‘It is great that Sina is tall.’)
b. E sili [lona maualuga].
 SILI the+.3. height
‘It is the highest.’ (Lit. ‘Its height is great.’)

The modest goal of this chapter will be to assign a semantics to sili that allows for a
unified analysis of both (5) and (6). Doing so will also shed light on the different
compositional routes that languages take to the construction of weak necessity,
priority modal meanings and to the construction of superlative meanings. We are
going to analyze sili as a gradable, degree-based predicate. This predicate is type-
polymorphic in that its underlying measure function may map both entities and
propositions to degrees. This measure function is partially under-specified and
dependent on context. In the morphologically unmarked form that we see in (5)
and (6), the positive may be pragmatically strengthened to a superlative-like
interpretation by recourse to alternatives, an interpretative option that is generally
available for gradable predicates in Samoan: In (7), umi ‘tall’ may both be
interpreted as a norm-related positive, under which Sina’s height is above the
contextual threshold for being tall. Under certain linguistic and contextual con-
ditions, it may, however, also receive a superlative-like interpretation (see also
Hohaus 2015).

(7) E umi Sina.

 tall Sina
‘Sina is the tallest.’ (Lit. ‘Sina is tall.’)

We are going to develop the analysis in the two steps: We will first discuss the
lexical semantics of sili in Section 2.2, and then the composition underlying the
examples above in Section 2.3. Our starting point in Section 2.2 will be the modal
OUP CORRECTED PROOF – FINAL, 16/6/2021, SPi

14  

readings of sili. We show that while sili syntactically patterns with non-epistemic
modal verbs in Samoan, it cannot plausibly receive a modal-quantificational
analysis. Unlike these verbs, sili can occur in a range of comparison constructions,
which rules out such an analysis (and is also is incompatible with hard-wiring a
superlative meaning into its lexical entry). Against the backdrop of the semantics
of gradable predicates in Samoan, we then develop a degree-based, relational
lexical entry for sili that also takes into account its non-modal readings.
Section 2.3 puts this lexical entry to use in the analysis of the comparative and
the morphologically unmarked form of sili. In the latter, the grammar of Samoan
exploits a general pragmatic option of strengthening the positive form to a
superlative-like interpretation through pragmatic reasoning or the exhaustifica-
tion of focus alternatives. We will first review this feature of the grammar of
Samoan, before we return to the analysis of the relevant examples with sili.
Section 2.4 concludes with a summary and with some desiderata for further
cross-linguistic research.
In the context of the current volume, this chapter is situated at the interface
between syntax and semantics, just like James N. Collins’ contribution in
Chapter 3. It is also situated at the interface between semantics and pragmatics.
Jens Hopperdietzel in his analysis of the Samoan prefix faʻa- in Chapter 6 of this
volume discusses another case of context-dependency in the grammar of Samoan,
contextually-governed allomorphy.

2.2 A lexical entry for sili

This section focuses on the lexical semantics of sili. Our starting point are those
examples which receive a modal interpretation. The examples initially appear to
be compatible with a quantificational-modal analysis. We will argue against such
an analysis and in favor of a relational-degree analysis based on the comparison
constructions in which sili can occur. This argument relies on an analysis of
gradable predicates in Samoan as making reference to degrees, and we will briefly
review the evidence in favor of such an analysis.

2.2.1 The modal interpretation

The verb sili is frequently used in the translation of imperatives as well as of

the weak necessity modal auxiliaries ought and should (see also Mosel &
Hovdhaugen 1992; Mosel & Soʻo 1997). Like those English modals (see also
von Fintel & Iatridou 2008), it is used for necessities. Unlike those modals, it is
restricted to the expression of priority modality (relating to permissions, goals,
desires, and the like), as shown in (5) in Section 2.1, and in (8) to (12). In all of
 OUP CORRECTED PROOF – FINAL, 16/6/2021, SPi

   15

these examples, the complement of sili is taken to describe the best course of
action out of the ones under consideration, and the sentences are used to express a
preference or give a recommendation. As such, they are compatible with con-
tinuations like the ones in (11c) and (12c), which flag that it is not an absolute
necessity that this course of action be taken.

(8) E sili [ona ʻe alu loa] ina neʻi tuai.

 SILI that you go then because lest late
‘You should go now, or you will be late.’
(Lit. ‘It is great that you go then, lest you are late.’) (Hunkin 1992: 49)

(9) E sili ai [loʻu moe i loimata maligi].

 SILI  my sleep  tear pour
‘Then, I’d rather sleep in tears.’
(Lit. ‘Then, my sleep in pouring tears is great.’)2

(10) E sili [ona oʻu fesili i le fomaʻi].

 SILI that I ask  the doctor
‘I should ask the doctor.’ (Lit. ‘It is great that I ask the doctor.’)

(11) a. [Giving advice to my friend who is learning how to drive:]

b. E sili [ona siaki muamua paʻu o lau
 great that check first tire of your
taʻavale aʻo leʻi malaga].
car while not+yet journey
‘You should check your tires before you start driving.’
(Lit. ‘It is great that you first check the tires of your car while
not yet traveling.’)
c. E leʻo se tulafono.
 not+ a law
‘It is not a law.’

(12) a. Context: Your friend Sina is visiting, but one of your sisters is about to
return; she doesn’t like Sina. You tell Sina:
b. E sili [ona e alu loa].
 great that you go then
‘You should go then.’ (Lit. ‘It is great that you go then.’)

² Veronica Leasiolagi Barber (1979), Tusi o Māfaufauga o Maona [Moana’s Diary], Honolulu:
Pacific Resources for Education and Learning.
OUP CORRECTED PROOF – FINAL, 16/6/2021, SPi

16  

c. Ae a leai, e leai se mea o iai.

but if    a thing  exist
‘But if not, it is not a thing.’

The verb is, however, not compatible with an epistemic interpretation, as is shown
in (13). In fact, as far as we are aware, none of the modal verbs in Samoan allow for
epistemic readings; epistemic modality appears to be encoded in sentence-initial
particles.³

(13) a. Context: You’re walking through your village with your sister. You
wonder whether your friend Sina and her family are at home. There is
music blasting from their fale, and the light is on. You conclude:
b. #Ua sili [ona i le fale Sina].
(inch) SILI that  the house Sina
‘Sina should be at home.’
(Lit. ‘It is now great that Sina is in the fale.’)4

Superficially, Samoan sili semantically and syntactically patterns with a whole

class of modal verbs, most prominently with the possibility modal verb mafai
‘to be possible’ and the necessity modal verb tatau ‘to be necessary’ in (14).
Adopting a standard quantificational approach for all of these modal verbs thus
suggests itself.

(14) E mafai [ona alu Rosa i le tifaga ma ana

 possible that go Rosa to the cinema with .3
uo], ae e tatau [ona toʻe foʻi Rosa i le
friend but  necessary that return again Rosa to the
fale i le tā o le valu].
house prep the beat of the eight
ʻRosa may go to the cinema with her friends, but she must be home by eight
o’clock.’

However, unlike the possibility modal verb mafai and the necessity modal verb
tatau, sili is gradable, that is, it can occur in a whole range of comparison
constructions.

³ A more systematic syntactic and semantic analysis of the grammar of modality in Samoan is yet to
be done, but see Mosel & Hovdhaugen (1992) for a descriptive overview.
⁴ Following up on a question from one of the reviewers: In the literal translation of this example,
‘now’ is intended to reflect the semantic contribution of the inchoative aspect.
 OUP CORRECTED PROOF – FINAL, 16/6/2021, SPi

   17

2.2.2 In favor of a degree-based analysis

As shown below, the verb sili is compatible with comparative and superlative
morphology as well as with degree intensifiers like the adverbial lava ‘really, very’.
In that respect, it patterns with other gradable verbs like umi ‘tall’ in (15a) and
(17a), and with mamao ‘far’ in (18a). The possibility modal verb mafai and the
necessity modal verb tatau are not gradable, however: Compare the minimal pairs
consisting of (15d), and (16a) and (16b), respectively.

(15) a. E umi atu Malia [nai lō Ioane].

 tall more Mary from  John
‘Compared to John, Mary is taller.’
b. [Discussing strategies for de-escalating conflicts among neighbors:]
c. [Talanoa ʻi le tagata] e sili atu [i lō le tusi ʻi ai.]
talk to the man  SILI more   the write to 
‘You should talk to the man rather than write to him.’
(Lit. ‘Talking to the man is greater compared to the writing to him.’)
d. E sili atu [ona fafaga e Me le pepe]
 SILI more that feed  Me the baby
[nai lō le tā o lāvalava].
from  the wash of clothing
‘Me should feed the baby rather than do the laundry.’
(Lit. ‘It is greater that Me feed the baby compared to the washing
of the clothes.’)
e. E sili atu [ona ʻe savali] [nai loʻo le ave o le taavale].
 SILI more that you walk from  the drive of the car
‘You should walk rather than drive.’
(Lit. ‘It is greater that you walk compared to the driving of the car.’)
f. E sili atu [ona ʻe vili iā Pati e fesili iai].
 great more that you ring to Patrick to ask +
‘It would be better if you called Patrick to ask him.’
(Lit. ‘It is greater that you ring Patrick to ask him.’)

(16) a. *E tatau atu ona fafaga e Me le pepe . . .

 necessary more that feed  Me the baby
Intended: ‘Me needs to feed the baby more than she needs to do the
laundry.’
b. *E mafai atu ona fafaga e Me le pepe . . .
 possible more that feed  Me the baby
Intended: ‘It is more possible for Me to feed the baby than to do the
laundry.’
OUP CORRECTED PROOF – FINAL, 16/6/2021, SPi

18  

%
(17) a. E aupito umi Malia.
  tall Mary
‘Mary is the tallest.’
b. [From a manual for teachers:]
Manatua, [o le mea aupito sili [ona e faamuamua]]
remember  the thing  SILI that you +
e le o le taumafai lea e aoao
 not  the attempt .  educate
atu mea uma.
 thing all
‘Remember, what you should prioritize is not to try to teach everything.’
(Lit. ‘ . . . the thing which it is greatest that you cause to be first is . . . ’)5

(18) a. Sā fotu mai mamao lava le vaʻa.

(.) appear from far  the ship
‘The ship appeared very far away.’ (Milner 1966: 102)
b. [From a textbook on Samoan culture:]
Ioe, e sili lava ona ʻe tapuaʻi.
yes  SILI  that you receive.support()
‘Yes, we should really support you (while preparing and giving that
speech).’ (Lit. ‘Yes, it is very great that you receive support.’)6

The compatibility with such morphology alone does not yet make the case for a
degree-based analysis, however. It has to be independently established that com-
parison constructions in the language require such an analysis (Beck et al. 2009;
Bochnak 2015; Deal & Hohaus 2019; Hohaus & Bochnak 2020); we briefly review
the evidence for Samoan below.

2.2.3 The semantics of gradable predicates in Samoan

Most comparison constructions can in principle also be analyzed with a vague

delineation semantics (Klein 1980; van Rooij 2011a, 2011b) that does not make
recourse to degrees but rather relies on manipulating the different partitions
induced by the gradable predicate. Whether gradable predicates are vague

⁵ Simon Dewey (2009), Tusi Lesona a le Faiaoga o le Tusi a Mamona [Book of Mormon Teacher
Manual], Salt Lake City: Religion Curriculum Services, Folasaga [Introduction].
⁶ Aumua Mataitusi Simanu (2002), O Si Manu a Alii: A Text for the Advanced Study of Samoan
Language and Culture, Honolulu: University of Hawai’i Press, p. 152.
 OUP CORRECTED PROOF – FINAL, 16/6/2021, SPi

   19

in this sense or degreeful appears to be a systematic point of variation in the

grammar of natural languages (Beck et al. 2009; Bochnak 2015; Bowler 2016;
Deal & Hohaus 2019).
This variation is captured by the Degree Semantics Parameter in (19).

(19) D S P [ DSP]: (Beck et al. 2009: 19, no.(62))
A language {does/ does not} have gradable predicates (type 〈d,〈e,t〉〉 and
related), i.e., lexical items that introduce degree arguments.

A negative setting of this parameter is, however, not appropriate for contemporary
Samoan: Following the line of argumentation in von Stechow (1984) and Beck
et al. (2009), Hohaus (2012, 2015) shows that a degree-based analysis of gradable
verbs and of comparative atu ‘more’ is needed to account for comparatives that
specify a differential measure, like in (20), and for comparatives where the
standard of the comparison is a degree measure, like in (21).

(20) E umi atu Malia [i le lua inisi] i lō

 tall more Mary  the two inch  
lona uso.
the+.3 sister
‘Mary is two inches taller compared to her sister.’

(21) E umi atu Mali [i lō le lima futu].

 tall more Mary   the five foot
‘Mary is taller than 5 ft.’

Under such an analysis, a gradable verb like umi ‘tall’ has the relational meaning
in (22).⁷
pffiffiffiffiffiffiffiffi
(22) ⟦ umi ‘tall’⟧ ¼ λw:λd:λx:(w)(x) ≥ d (type 〈s; 〈d; 〈e; t〉〉〉)

Given that sili patterns with other gradable predicates in the language and given
that gradable predicates in Samoan have a degree-based semantics, we conclude
that sili should be analyzed as a degree predicate, rather than under the standard
quantificational analysis of modal expressions.

⁷ Interestingly, this is a recent development in the language from Early Written Samoan
(1850–1900) to Present-Day Samoan, which Hohaus (2018) argues had a negative setting of the
DSP.
OUP CORRECTED PROOF – FINAL, 16/6/2021, SPi

20  

2.2.4 The semantic type of sili

While the gradable verb sili has in common with a gradable verb like umi ‘tall’ that
its semantics is degree-based, it is different from umi ‘tall’ in that it can combine
with both individual- and proposition-type arguments. It therefore has to be
analyzed as type-polymorphic, that is, of type 〈s,〈d,〈α,t〉〉〉, where α can be either
of type e or of type 〈s,t〉. It shares this polymorphism with several other gradable
predicates like, for instance, lelei ‘good’ in (23), and with its English counterpart
(Villalta 2008; Gergel 2011; Lassiter 2017), in (24).

(23) a. Ua lelei le matagi.

(inch) good the wind
‘The wind is good.’
b. E lelei ʻona alu ʻese ʻoia.
 good that go away +.3
‘It would be good if he/she were to go away.’
(Lit. ‘It is good that he/she goes away.’) (Chung 1978: 167, no.(158b))

(24) a. This recipe is good.

b. It is good that Bill is here. (Lassiter 2017: 175, no.(7.1))

If the lexical semantics of sili relies on mappings from individuals and proposition
to degrees, how can we characterize the measure function that is at the core of its
lexical entry? We take up this question next.

2.2.5 The underlying measure function

In the case of the modal readings discussed in this section so far, the underlying
scale is plausibly one of modal priority, relating to desirability, preference, or
usefulness towards a goal. These measures are highly dependent on the linguistic
and situational context of an utterance. They can, however, be related back to the
notion of accessibility relation and of ordering source that feature prominently in
the literature on modality, as discussed by Portner & Rubinstein (2016; see also
Klecha 2014 and Pasternak 2016): Under the quantificational analysis, the domain
of quantification of a modal expression are those worlds accessible through the
accessible relation that are highest ranked with respect to one or—in the case of
weak necessity modals (von Fintel & Iatridou 2008; Rubinstein 2012, 2014)—two
ordering sources. The perceived weakness of should, illustrated in (25), under this
view is due to this second ordering source, which additionally restricts the domain
of quantification of the universal quantifier, resulting in the a weaker statement.
Rubinstein (2014) suggests that the propositions in the second ordering source
 OUP CORRECTED PROOF – FINAL, 16/6/2021, SPi

   21

differ from the ones in the primary ordering source in the level of commitment the
speaker has to them; they can be characterized as negotiable priorities. As the
exchange in (25) shows, the ambassador’s recommendation is based on considera-
tions that are clearly not shared by the author of the post (say, controlling the
news cycle or being consulted during the writing process).

(25) [From an exchange on social media:]

Post: US Ambassador to Berlin Richard Grenell writes letter to Spiegel
magazine suggesting journalist should call US Embassy before they start
writing. I dunno.
Reply: They should, but don’t have to.8

Portner & Rubinstein (2016) propose that the propositions in the secondary
ordering source can be used to define a degree scale (and an associated measure
of modal necessity) based on the relative importance the speaker assigns to those
propositions. To this end, they exploit that degrees can be reconstructed from
equivalence classes, sets of items that are indistinguishable from each other with
respect to a certain property (for instance, the importance the speaker assigns to
them). If these sets are totally ordered, we have a scale. Assuming that context
ranks all possible non-empty subsets of the secondary ordering source by their
relative importance, we can construct a a context- and world-dependent degree
scale for propositions Sw;c as in (26). Degrees are here equivalence classes of
propositions, which are compatible with the circumstantial modal base and best
with respect to the first and these ranked subsets from the second ordering source
ord′. In (26), Rc is a contextually provided function that induces an importance
ranking on the subsets in the secondary ordering source. Measuring a proposition
against this scale ultimately relates it to how well the proposition fares with respect
to our priorities in the secondary ordering source.

(26) For any world of evaluation w and context c: Sw;c ¼ 〈Dw ; >〉, where
Dw ¼ ffp ∈ D〈s;t〉: ord1ðwÞ;ord′ ð∩ accessðwÞÞ  p}: ord′ ∈Rc ðord2ðwÞÞ},
and, for any two degrees d and d′ ∈ Dw ; d > d′ if and only if d ′ ⊂d.
(adapted from Portner & Rubinstein 2016: 275, no.(43))

Hard-wiring a measure of modal priority based on (26) into the lexical meaning of
sili is, however, not appropriate in the light of the other readings that it allows for.
Rather, a measure function based on (26) is one of several measures that context
can make available and with which sili is compatible. The division of labor
between world knowledge, context, and lexical meaning in determining the

⁸ Twitter exchange between users @ConStelz and @mtavkhelidze on December 22, 2018. https://
twitter.com/mtavkhelidze/status/1076574419781332993 (accessed Dec. 30, 2019).
OUP CORRECTED PROOF – FINAL, 16/6/2021, SPi

22  

measure used in a particular situation for the interpretation of sili most certainly
deserves further attention. For the purposes of this chapter (and the questions of
compositionality that it attempts to answer in Section 2.3), it will be sufficient to
analyze sili as in (27). It is based on a lexically partially under-specified and thus
context-dependent measure of  of its argument. In that respect, it is most
likely similar to its English counterpart great.
pffiffiffiffiffiffi
(27) ⟦ sili⟧c ¼ λw:λα:λd:c(w)(α)≥ d
with α of type e or of type 〈s,t〉

An analysis along the lines of (27) will allow sili to receive non-modal interpreta-
tions as well, to which we turn next.

2.2.6 Other readings

As we have already seen briefly in the Section 2.1, sili may also be used to compare
other types of values that can be associated with an entity or proposition. Further
examples are below. The examples in (28) to (30) are based on a value or extent
relating to a property concept like height, weight, and length: The value assigned
to a noun phrase like le mamafa o Ioane ‘John’s weight’ in (28), for instance, will
be a degree associated with a standard unit of measurement for weight, like stone,
kilogram, or pound. For cocoa or lifestyle, we lack such standardized units of
measurements, and the comparison in (31) and (32) is based on a somewhat more
subjective measure of quality or worth.

(28) E sili atu [le mamafa o Ioane] [i lō Malia].

 SILI more the weight of John   Mary
‘John weighs more than Mary.’
(Lit. ‘Compared to Mary, John’s weight is greater.’)

(29) a. [Talking about Mount Silisili, the highest mountain in Samoa:]

b. E sili [lona maualuga].
 SILI the+.3 height
‘It is the highest.’ (Lit. ‘Its height is great.’)

(30) O le Salamo lenei e aupito sili lona umi.

 the psalm .   SILI the+.3 length
‘This psalm, it is the longest.’ (Lit. ‘This psalm, its length is the greatest.’)
 OUP CORRECTED PROOF – FINAL, 16/6/2021, SPi

   23

(31) E sili [le koko Samoa].

 SILI the cocoa Samoan
‘Samoan cocoa is the best.’ (Lit. ‘Samoan cocoa is great.’)

(32) E sili [lava le faasamoa]!

 SILI  the custom+Samoa
‘The Samoan way of life is the very best!’
(Lit. ‘The Samoan way of life is very great.’)

What about those examples where sili takes an ona-complement but does not
receive a modal interpretation? This is the case if the complement contains a
gradable predicate as in (33), for instance. While the example receives an inter-
pretation that can be paraphrased by a superlative (= ‘Frank is the tallest’;
see Section 2.3), we do not think it is plausible that the embedded proposition
will be measured along the scale that the embedded gradable predicate is based
on, that is, sili in (23) does not relate the proposition that Frank is tall to
Frank’s height degree. We suggest that these types of examples talk about the
meta-linguistic appropriateness or fit of the proposition in the context, that is,
 will be a measure of contextual fit, and thus a measure of how appropriate
it is in the context to say that Frank should be considered tall, considering
other things that the speaker could have said (that is, that Mary is tall or that
John is tall).

(33) a. Context: Mary’s height is 1.78 meters, John’s 1.82 meters, and Frank’s
1.86 meters.
b. E sili [ona umi Falani].
 SILI that tall Frank
‘Frank is the tallest.’ (Lit. ‘It is great that Frank is tall.’)
c. ʻO Falani e sili [ona umi].
 Frank  SILI that tall
‘Frank is the tallest.’ (Lit. ‘Frank is such that it is great that he is tall.’)

Additional examples with the gradable verbs fefe ‘to fear’ and fiafia ‘to like’ are in
(34) and (35). Comparative- and superlative-marked occurrences of this use of sili
are also attested; examples are in (36) and (37). These examples show that the
superlative-like interpretation that is available for sili must be compositionally
derived. We turn to the how in the next section.

(34) ʻO ai e sili [ona ʻe fefe iai]?

   SILI that you fear +
‘Whom do you fear most?’
(Lit. ‘Who is such that it is great that you fear?’)
OUP CORRECTED PROOF – FINAL, 16/6/2021, SPi

24  

(35) ʻO le ā le aega o le tala e sili

 the what the section of the story  SILI
[ona ʻe fiafia i ai]?
that you like  
‘Which part of the story did you like best?’
(Lit. ‘What is the part of the story such that it is great that you like it?’)

(36) O le mea e sili atu [ona leaga [i lo

 the thing  SILI more that bad  
le pomu ʻatomika]]?
the bomb atomic
‘What is worse than an atomic bomb?’
(Lit. ‘The thing that is such that it is greater that it is
bad compared to the atomic bomb?’9)

(37) O le masina o Me 17th 1881 na lomia ai

 the month of May 17th 1881 (.) publish 
ile nusipepa o le Chicago Times
+the newspaper  the Chicago Times
[se feau faauaealesi aupito sili [ona umi ile
a cable message  SILI that long +the
tala faasolopito o le lalolagi, e 118,000 upu]].
history of the world  118,000 word
‘On the 17th of May 1881, the Chicago Times published a cable message,
which with 118,000 words was the longest in the history of the world.’10

2.2.7 Interim summary

We have shown that the Samoan verb sili is gradable and occurs in a variety of
degree-based constructions. Given that Samoan can be characterized as having a
positive setting of Beck et al. (2009)’s Degree Semantics Parameter, we have used
these data to argue for a degree-based semantics of sili, rather than a modal-
quantificational analysis. Sili is under-specified and thus context-dependent as to
the dimension of value measured, but allows for measures relating to weak modal
necessity, to property concepts and to contextual appropriateness, depending on
the linguistic and extra-linguistic context. In the next section, we will discuss how

⁹ Alaiseā F. Iosefa & Veronica Leasiolagi Barber (2012), Riddles: Supplementary Pages for Samoan
Resource Book. Honolulu: Pacific Resources for Education and Learning.
¹⁰ Ekalesia Faapotopotoga Kerisiano Samoa [Congregational Christian Church Samoa] (2015), Tusi
Aʻoga Mo Autalavou [Textbook for Youth], Apia: Matagaluega o Aʻoaʻoga Kerisiano [Department of
Christian Education], p. 6.
 OUP CORRECTED PROOF – FINAL, 16/6/2021, SPi

   25

this lexical meaning compositionally interacts with different degree operators,

focus, and pragmatic reasoning.

2.3 Composition

In this section, we will look at questions of composition, and more specifically at

what happens to the degree argument that sili introduces. We will discuss two
cases, the interpretation of examples where sili combines with comparative mor-
phology and the interpretation of examples with the morphologically unmarked
form, for which we have observed a mismatch between the positive form and its
superlative-like interpretation.

2.3.1 The interpretation of comparatives with sili

In the case of a comparative, the degree argument position that sili introduces
syntactically is filled by the comparative operator. We discuss here the analysis of
(38), a so-called contextual comparative where the standard of the comparison is
implicit.¹¹ In the context provided, comparison is between buying a random
selection of pet food for the mystery pet or calling Patrick to ask him to reveal its
identity.¹² Stacey’s friend expresses a preference for the latter over the former.
Truth-conditionally, this preference translates to (39) under a degree-based analysis.

(38) a. [Patrick asked his friend Stacey to pet-sit Fluffy while he is on vacation.
Stacey accepted but realizes that she has forgotten to ask what animal
Fluffy is when she is at the store shopping for pet food with a friend.
They consider just buying a selection of pet food or calling Patrick.]
b. E sili atu [ona ʻe vili iā Pati (e fesili iai)].
 great more that you ring to Patrick to ask +
‘It would be better if you call Patrick to ask him.’
(Lit. ‘It is greater that you ring Patrick to ask him.’)
(39) Truth conditions: c (w@)(p) > dc
where w@ is the actual world, p = [λw.  rings P in w]
and dc = the degree that is assigned to the alternative proposition in w@

¹¹ We thereby set aside questions relating to the syntax of the standard phrase in the Samoan
comparative, which Hohaus (2015) argues to be phrasal but not an argument of the comparative
operator (and, rather, a frame-setting adverbial). These questions are orthogonal to the plot of the
chapter.
¹² From follow-up elicitation based on the TFS Working Group (2012)’s storyboard ‘Feeding Fluffy
Totem Field Storyboards http://totemfieldstoryboards.org/stories/feeding_fluffy/ (accessed Dec. 29,
2019).
OUP CORRECTED PROOF – FINAL, 16/6/2021, SPi

26  

The composition relies on a type-polymorphic variant of the comparative

operator, in (40), and proceeds as sketched in (41).

(40) ⟦atu ‘more’⟧c ¼ λR〈d,〈α,t〉〉.λα.(λd.R(d)(α)=1)>dc

where α is either of type e or 〈s,t〉
(adapted from Hohaus 2015: 131, no.(258))

(41) Composition:
For any context c, ⟦atu⟧c ð⟦sili⟧c Þð⟦½ona:::⟧c Þ ¼ 1 iff
[λR〈d,〈α,t〉〉.[λp〈s,t〉.(λd.R(d)(p)=1)>dc]]
([λp〈s,t〉.[λd.c(w@)(p)≥d]])([[[ona . . . ]]]c) = 1 iff
(λd.c(w@)([[[ona . . . ]]]c)≥d) >dc iff
c(w@)([[ona . . . ]]]c) >dc

The measure function  here presumably is contextually fleshed out as a

measure constructed from any additional priorities her friend assumes Stacey
has in that situation, plausibly including not embarrassing herself, not spending
money on pet food for the wrong animal, getting the shopping done as quickly
as possible, etc. The more of those priorities a proposition meets, the higher
its value.

2.3.2 The interpretation of the positive form

In the absence of an overt degree head like comparative atu, the degree argument
that the gradable predicate makes available will be bound off by a covert Positive-
operator, as is standardly assumed in degree semantics (Kennedy 1997, 2007; Fara
2000; Fults 2006; von Stechow 2009; Beck 2011). When it comes to the interpre-
tation of the morphologically unmarked form of gradable predicates, Samoan,
however, not only allows for the norm-related and vague interpretation that we
find in English, but also for a superlative interpretation (Hohaus 2015). This
interpretative behavior requires an excursus, after which we will return to the
examples with sili.
In out-of-the blue contexts like (42), the sentence with tele ‘many’ is true if and
only if the number of cars in Samoa is now such that the speakers considers them
to be many, that is, within a vague norm for what counts as many cars in the
context of a small island state. The examples in (43) and (44), however, where
the elicitation context implicitly sets up a comparison (Who’s taller, John or
Mary? Who’s older than whom?), receive a superlative interpretation by default.
In (43), consultants consistently reject the statement that Mary is tall, even
though she clearly is at 182 centimeters. In (44), while we would be reluctant
in English to describe any of these children as old, the sentence that Sina is old is
 OUP CORRECTED PROOF – FINAL, 16/6/2021, SPi

   27

judged true. As one of our language consultant’s explanation in (44c) shows, the
sentence is judged true because Sina is the oldest of the three children.
Frequently these interpretations are accompanied by focus movement of the
subject noun phrase, as in (44d).

(42) a. Context: Imagine your sister hasn’t been back to Samoa in 35 years. She is
surprised at the number of cars there are now.
b. Ua tele taʻavale i Sāmoa!
(inch) many car in Samoa
‘There are now many cars in Samoa!’

(43) a. Context: Mary’s height is 6 feet. John’s height is 6 feet 3 inches.

b. #E umi Malia.
 tall Mary
‘Mary is the tallest.’ (Lit. ‘Mary is tall.’)
c. E umi Ioane.
 tall John
‘John is the tallest.’
(Lit. ‘John is tall.’) (Hohaus 2015: 128, no.(243)–(244))

(44) a. Context: Imagine that your niece Malia is six years old, your nephew Tasi
is four years old, and your daughter Sina is seven years old.
b. E matua Sina.
 old Sina
‘Sina is the oldest.’ (Lit. ‘Sina is old.’)
c. E leʻo Tasi e matua, e leʻo Malia foʻi, ae ʻo Sina.
 + Tasi  old  not+ Malia also but  Sina
‘It is not Tasi who is old, it is also not Mary, but it is Sina.’
d. ʻO Sina e matua.
 Sina  old
‘It is Sina who is the oldest.’ (Lit. ‘It is Sina who is old.’)
[[  NP ]α [  verb __α ] ]
↑_ _ _ _ _ _ _ _ _ _ _j

Rather than positing two covert operators to derive these different readings (one
introducing norm-relatedness and vagueness, the other a proper superlative), we
suggest that the latter reading is ultimately pragmatic and relies on a manipulation
of what it means to be tall or old in a certain utterance context plus strengthening
(through pragmatic reasoning or through the exhaustification that the interpre-
tation of focus triggers). This type of semantics also underlies the so-called
conjoined comparative construction that Samoan makes use of (Marsack 1975;
OUP CORRECTED PROOF – FINAL, 16/6/2021, SPi

28  

Stassen 1985; Hohaus 2018); an example is in (45). This example indirectly

brings about a comparison (John’s age > Mary’s age) by classifying John as
old for the purposes of the utterance and Mary as young (see also Hohaus &
Bochnak 2020). Again, we would usually not independently consider a five-year
old to be old.

(45) a. [Playmobil figurines, two parents and two children. Mary’s age is given as
three years, John’s as five years.]
b. [E matua Ioane] [ae laitiiti Malia].
 old John but young Mary
‘John is older than Mary.’ (Lit. ‘John is old, but Mary is young.’)

More technically, the interpretation of the positive form in Samoan relies on the
operator in (46). This -operator requires that the maximal degree that its
argument reaches lies within a certain part of the scale, building on the idea of a
tri-partite structure of degree scales. Just like the operator proposed in Stechow
(2009), (46) is based on the assumption that any degree scale S (like the one for
height, for instance) “is divided into three parts, a part that contains the short
things under consideration, a part that contains the tall things under considera-
tion, and a neutral part that contains the things that are neither short nor tall”
(von Stechow 2009: 220).

(46) [[]]c = λR〈d,〈α,t〉〉.λα.(λd.R(d)(α)=1)∈ S+R,c

with α of type e or type 〈s,t〉

(47)
R
S – neutral +
R,c R,c R,c

The -operator assigns the maximal degree that its second arguments reaches to
the upper part of the scale Sþ R;c , which is based on the things we consider to be
tall or old or many in the utterance context. Applied to our example in (42) above
(= ‘There are now many cars in Samoa’): Saying that there are many cars in Samoa
places the current number of cars on the islands in the upper part of the scale,
within Sþ þ
CARDINALITY;c . Which section of the scale is going to constitute SCARDINALITY;c
is ultimately context-dependent, and will remain vague in most cases. The
tri-partition of the scale can, however, also be manipulated overtly, as is the case
in the conjoined comparative. In (45) above (= ‘John is old but Mary is young’),
these partitions are made, for the purposes of the comparison, on the basis of
Mary and John’s age only. If we assign John’s age to Sþ AGE;c and Mary’s age to
S AGE;c , it follows that John’s age exceeds Mary’s. It does, however, not follow that
Another random document with
no related content on Scribd:
avaimenreikään. Se oli hirveän raskas, mutta lukossa, ja avainta ei
näkynyt. Hän olisi tahtonut nähdä oliko Marin puheissa perää.

Hamberg oli aina hyvällä päällä kotona ollessaan; hän koetti

ilmeisesti haihduttaa Fransiljan ikävää lystillisillä jutuilla milloin
mistäkin kalastajasta tai tapahtumasta merellä tai kaupungin
kalarannassa. Hän kertoi niin hassuja että täytyi väkisinkin nauraa.
Ja väliin hän kujeili kuin pojat, kiusasi ja nipisti.

Kerran Hamberg kertoi Boströmistä. Boström oli kalatorppari ja

hänen veroonsa kuului kaksi kertaa viikossa käydä kaupungissa
talon maitoa ja piimää kaupalla. Eräänä yönä myöhään keväällä, kun
Hamberg oli illalla kokenut isotrysät ja lähtenyt kaupunkiin
saadakseen hyvän paikan laiturin kyljessä, oli hänen veneensä
joutunut rinnan Boströmin kanssa. Kumpikin nukkui purjeteltassa,
mutta kun Hamberg heräsi siihen että joukko humalaisia
valkolakkeja rallattaen kulki laiturilla, pisti hän päänsä purjeen alta ja
näki Boströmin varovasti ammentavan vettä merestä
kirnupiimäsaaviin. Yksi ylioppilaista, joka seisoi Koroleffin kulmalla oli
huomannut metkun. Hetken päästä tuli hän tahtomaan kaksi kannua
kirnupiimää itselleen ja tovereilleen, jotka odottivat kauempana.

— Onkos herralla jotain astiaa sitten, sanoi Boström.

— Ei ole. Pankaa nyt vaikka tuohon säkkiin.

— Säkkiin? Boström katseli epäillen ostajaan, mutta tämä kehoitti

vaan kaatamaan säkkiin ja tarjosi rahaa.

No mitäs muuta, Boström mittasi säkkiin ja ylioppilas kumartui

veneeseen, ottamaan säkkiä muka, mutta sieppasikin sen käteensä
ja alkoi sillä huimia Boströmiä korville.
 Vai merivettä sinä ajat saaviisi veijari! Kyllä minä opetan sinut sen
vietävä petkuttamaan ihmisiä piimälläsi! Semmoista kirnupiimää on
tämä allas täynnä! kirkui ylioppilas ja veteli ehtimän takaa säkillä. —
Laita luusi täältä tahi minä ilmoitan konstisi poliisille, hän räyhäsi ja
päästi Boströmin veneen renkaasta ja potkasi sen jalallaan rannasta.
Boström oli punainen kuin kukon heltta ja kun hän huomasi että yhä
useimmissa veneissä herättiin rähinään ja naurettiin hänelle, sousi
hän kiiruumman kautta tiehensä. Viikon päiviin ei näkynyt häntä
senjälkeen venerannassa, lopetti Hamberg.

Fransilja mieltyi pian oloonsa. Hän koetti tehdä työt hyvin

ollakseen ukolle mieliksi eikä välittänyt Marin juorupuheista. Eräänä
sunnuntai-aamuna toi hän Hambergille 25-pennisen, jonka oli
löytänyt pöydän alta. »Tyttö on rehellinen», ajatteli Hamberg
tyytyväisenä, sillä hän oli tahallaan pudottanut sinne rahan
koetellakseen Fransiljaa. Toisen kerran löysi hän ukon kintaan
kaapin alta. Senkin oli Hamberg paiskannut sinne tahallaan
saadakseen nähdä miten lavealti hän lakaisi lattiaa.

Näin kului syksy, tuvassa ja vesillä, paikkuupukin ja airojen

ääressä. Pyhäinmiesten jälkeen tulivat myrskyt, viimeiset rysät
otettiin maalle täynnänsä leviä ja ahvenruohoja, raakkuja ja
vesilimaa. Lehti oli jo aikoja sitten pudonnut puusta, koivut pieksivät
ilmaa paljailla varvuillaan, tuuli muuttui viluiseksi viuhinaksi, meri
kohisi raskaana ja synkkänä rantoihin.

Kun vettynyt verkkovene oli saatu parahiksi maalle, tuli

ensimäinen lumi rajun itä-tuulen tuomana. Yöllä rajuilma yltyi ja
viisari katossa keikkui levottomasti sinne tänne, kun vihurit
ravistelivat ja peruuttivat peltikukkoa katolla.
 Hamberg makasi valveillaan sängyssä. Hän suunnitteli uutta
venettään ja ajatteli huomena veistää siihen emäpuun. Mutta silloin
hän muisti että suurempi kirves oli kesästä saakka ollut kateissa eikä
vieläkään ollut löytynyt. Ja hän ihmetteli mihin se oli voinut joutua,
ellei se ollut unohtunut Skytteskäriin, silloin heinäntekomatkalla. Ja
harmitteli mielessään hyvää kirvestä.

Tänä yönä näki Hamberg Rosinan unissaan.

Hän tuli kuin olisi nukkunut tuvassa, lyhyessä paidassaan

Hambergin vuoteen luo.

Isä-kulta, etkö aio ottaa kirvestä sireenipensaasta, muuten se jää

lumen alle, hän sanoi ja katsoi huolestuneena mieheensä.

Hamberg kavahti istualleen ja hieroi silmiään. Oliko hän heteillään

vai nukuksissa? Ulkona vinkui myrsky, peltikukko äänteli ja
sireenipensas raapi ikkunaan. Mutta arvelematta sen enempää hän
kulki paljain jaloin poikki lattian ja tupaan, jossa Silja nukkui hupussa
ja avasi oven. Kello löi juuri kolme.

Ulkona tanssivat ja pyörivät lumihiutaleet ilmassa ja tukkivat silmät

hänen kumartuessaan pilkkopimeässä sireenipensasta kohti. Ensin
tuli käteen särkynyt aironlapa, mutta sitten, aivan oikein, kirves. Ja
kirves kädessä hän palasi tupaan, laski sen uunin eteen huomiseksi
ja nukkui jälleen kummastellen kuinka Rosina pitää kaikesta huolen.

Tämä ei ollut ensi kertaa kuin Rosina ilmestyi hänen avukseen.

Ensi kerran tapahtui se viime pääsiäisenä, kun Hambergin oli
mentävä rippikirkkoon niinkuin heillä oli ollut tapana Rosinan kanssa
käydä joka vuosi, eikä hän löytänyt mistään papinkirjaa. Hän oli
hakenut moneen kertaan kirstunsa ja kaappinsa ja aika oli täpärällä.
Silloin tuli Rosina eräänä yönä hänen luokseen sanoen: —
Papinkirjahan on peilin laatikossa. Ja hän oli vielä lisännyt: — Aja
parta nenäsi alta, muuten en tule sinun kanssasi! Ja kun Hamberg
niskotteli ja antoi viiksensä kasvaa, niinkuin hän oli antanut jälkeen
Rosinan kuoleman, tuli hän uudelleen ja sanoi: — Isä-kulta, aja pois
parta nenäsi alta, sinä olet kuin ryssän musikka! Ja silloin Hamberg
totteli.

Mutta aamulla tämän viime näyn jälkeen tahkoi Hamberg

kirveensä, se kun oli pensaassa pahasti ruostunut ja alkoi veistellä
emäpuuksi sitä männynkäyrää, jonka hän jo kesällä oli katsonut
itselleen Skytteskäristä.

*****

Tuli adventti hiljaisine päivineen, jolloin aurinko pitkästä aikaa

näyttäytyi saarelaisille, puhkaisten kekälepunaisena sumujoukot,
jotka täyttivät välin merestä taivaaseen. Vedet lepäsivät joka taholla
huuruten ilmaan viimeisen lämpönsä, ja odottivat kantta päälleen.

Eräänä yönä oli pakkanen vetänyt sinisen kalvon yli läntisen selän.
Kaukana taivaanrannalla näkyivät laineet viluisessa aamu-ilmassa
vielä kulkevan vapaina, kuin kiiruhtaen jotain kaukaista määrää
kohden. Vilkkumajakat sammuivat, ulkosaaret ja luodot yksinäistyivät
selille ja nyt tuli Aspöläisillä olla perunat ja ruissäkit aitassa, sillä
yhteys muihin ihmisiin oli katkaistu pitkäksi aikaa. Jää ei kantanut,
veneellä ei päässyt. Talvi oli tullut.

Toisen adventtisunnuntain viikolla, kun Hamberg ja Fransilja olivat

teurastaneet porsaan ja paraillaan kalttasivat harjaksia sen selästä,
tuli Fanny miehineen tervehtimään isäänsä. Hän oli laiha ja luiseva,
kuiva kuin kapakala. Äitinsä mustat silmät hän oli perinyt, mutta ääni
oli kimeä ja särkynyt alituisesta haukkumisesta. Mies oli juoppo
koneenkäyttäjä. Tyttönä oli Fanny kauan valikoinut sulhasia, pojat
olivat ruvenneet välttämään häntä hänen terävän kielensä takia ja
vanhanapiikana hän sitten kostoksi oli hakenut miehen itselleen
kaupungista, esiintyen saarelaisten rinnalla nyt rouvana. Fransilja
katsoi silmät, selällään hänen hattuaan, se oli kuin rastaanpesä. Ja
takissa oli koketit ja kamyyrit.

Fanny asetti suunsa mielistelevästä suppuun kun hän tervehti

isäänsä ja kysyi hänen vointiaan, mutta Fransiljaan hän iski
epäluuloisesti mustat silmänsä. Näytti kuin olisi hän tahtonut katsoa
läpi luiden ja ytimien.

Kun he tuvassa olivat kahden ja Fransilja sytytteli tulta pesään,

huomasi Fanny kauniin rintasoljen kaapin päällä.

— Keltä Fransilja on tuon saanut? kysyi hän terävästi.

— Sedältä, vastasi Silja punehtuen kuin olisi hänet tavattu

pahanteosta. Ja sitten pujahti hän katokseen puita hakemaan ja
viivytteli siellä siksi kunnes Hamberg tuli aitasta, jossa hän oli
ripustanut siankimpaleet kattoon. Hän pelkäsi olla kahden tuon
ihmisen kanssa.

Fanny oli syrjästäpäin kuullut että Hamberg oli ostanut tytölle

hamekankaan ja muitakin epäilyttäviä puheita ja hän oli tullut
vartavasten ottamaan asioista selon. Oliko todella mahdollista että
isä oli niin järjetön, niin vanhuuttaan höperö, että hänellä olisi ollut
jotain aikeita Fransiljan suhteen? Että tuo Köökarin vuohi olisi saanut
villityksi vanhan miehen! Että isä olisi niin kiittämätön kaikesta siitä
rakkaudesta, mitä hän ja Emil olivat aina osottaneet tälle! Kun hän
oikein ajatteli, oli se verinen vääryys, joka ei saisi koskaan tapahtua,
ei koskaan, ei koskaan!

Hamberg on tyytymätön tämänvuotiseen porsaaseen. »Senverran

on silavaa tuossa meidän Patussakin», hän sanoo kahvia
hörpätessään. Ja sitten hän Fransiljalle ja Emilille alkaa kertoa siitä
takavuosien jättiläissiasta, joka oli ollut semmoista lajia kuin olisi se
juossut kärsänsä lyttyyn seinää vasten. Mutta sekös otti ruokaa ja
turposi! Lopulta se vain röhötti paikoillaan ja ajankuluksi alkoi järsiä
lätin kattoa niin että sitä oli kohotettava yhdellä hirsikerralla. Siinä oli
sikaa! Hänen silmäteränsä laajenevat pelkästä hyvästämielestä,
muistellessa sitä jättiläissikaa.

Emil on heitellyt kaihoavia silmäyksiä seinäkaappiin ja vihdoin

nousee Hamberg ja tuo rakkopullon pöydälle. Pieni puolikupponen
kuuluu teurastuspäivän kestitykseen. Kun Fannyn silmä välttää
ryyppäsee Emil kahvikuppinsa melkein tyhjiin ja liruttaa sen sitten
pullosta taas täyteen. Hänen katseensa harhailee epävarmana pitkin
tupaa ja pysähtyy vihdoin katossa olevaan leiviskänpainoiseen
hyljeluodikkoon. Hän on kookas ja turpea mies, mutta pehmeätä
tekoa. Silmäluomet nousevat raskaasti ja katse niiden alla on
jäykästi tuijottava. Fannyyn hän välttää katsomasta.

Kun kuulumiset molemmin puolin on kyselty ja keskusteluun tullut

ensimäinen äänettömyys, nousee Fanny päättävästi ja sanoo
Hambergille:

— Jos isä tulisi vähän tänne kamariin, minulla on asiaa.

Fransilja kumartuu lieteen päin, silmissä entinen pelokas ilme.

Emil tuijottaa yhä luodikkoon, nyt hänen täytyy ottaa se katosta alas,
punnita käsissään ja ikkunan ääressä tarkastella sen piilukkoa.
 — Onpas siinä luodikkoa! Mutta kuka tämmöistä jaksaa kädellään
kannattaa?

Fransilja naurahtaa miehen tyhmyydelle meriasioissa, selittäen

että sillä ammutaan vain jään nojalta tai kiveltä. Mies makaa
vatsallaan ja tähtää sen hylkeeseen.

Joku uusi ajatus alkaa liikkua Emilin hitaassa päässä: hän

silmäsee varovasti että kamarin ovi on kiinni ja antaa katseensa
sitten mielihyvällä liukua pitkin Fransiljan olkapäitä ja rintoja.

Ei hullumpaa, ei yhtään hullumpaa! Ukko kyllä tietää mitä tekee.

Hän uskaltaa nyt vasta oikein katsoa Fransiljaa, hyväillä silmillään

hänen verevää muotoaan ja punaisia huuliaan. Turkanen kuinka ne
palavat!

Hän muuttuu huomaavammaksi, koettaa saada ryhtiä itseensä,

pitää liikkeensä tasapainossa. Jos pikku Fransilja antaisi hänelle
uutta kahvia! Kuinka pikku Fransilja viihtyy täällä autiolla meren
saarella, jossa ei ole muuta seuraa kuin… kuin…

Hän nauraa silmät kiltissä vanhan juopon mielistelevää naurua,

laskee leikkiä sulhasista, maiskahuttaa huuliaan.

Silloin viiltää hänen korvaansa tuttu ääni, terävä kuin partaveitsi.

— Jos ei isä anna, istun tässä kirstun päällä, siksi kun se

mätänee!

Ääni tulee kamarista. Nyt se kirahtaa kuin sahan osuessa pahasti

puuhun. Ovesta kuuluu lyhyt käsky.
 — Emil, tule tänne!

Emililtä haihtuvat siinä tuokiossa kaikki keikarieleet, silmät

räpyttävät epävarmoina siitä kuulivatko korvat oikein. Hänen sameat
aivonsa tapaavat jostain toisen ajatuksen: matkan tarkoituksen,
perinnöt jotka voivat mennä menojaan tuon tyttöhepsakan tähden ja
siihen ajatukseen hän julmistuu. Mutta jo ovessa lennättää Hamberg
hänelle vastaan:

— Sinulla ei tässä ole mitään sanomista! Jos olisit seissyt edes

vihillä omissa housuissasi!

Emil peräytyy kynnykseltä kuin olisi saanut vettä silmilleen.

Fannyn oli täytynyt Hambergilta lainata mustat housut heidän vihille
mennessään.

Kamarissa istuu Fanny Hambergin rahakirstun päällä, kädet

sivuillaan ja harmin puna poskilla. Ukko on tuikean ja ikävystyneen
näköinen. Hänen huulensa liikkuvat kuin pureskelisi hän jotain
karvasta.

— Jos isä vähänkin ajattelusi, jos isä tahtoisi kuulla… alkoi Fanny
uudelleen, tällä kertaa vetäen toisesta köydenpäästä. Isälle se ei
paljo tunnu, mutta meille ne kaksituhatta ovat ihan välttämättömät.
Velka lankee, osake on maksettava tammikuussa…

Emil säestää ruumiillaan ja eleillään hänen sanojaan; nyt hän tohtii

aukaista suunsakin:

— Isä kulta, kuinka isä voi…

Mutta Hamberg katkaisee lyhyeen:

 — Sinä! Hakisit korkoja olutpatruunalta, jolle olet kaksikymmentä
vuotta tallettanut palkkaasi!

Emilin suu menee umpeen, silmät alkavat taas harhailla, kunnes

kaapin lasin takaa tapaavat suuren hopeamitalin »ihmishengen
pelastamisesta» punaisine kaulanauhoineen.

Mutta Hamberg on käynyt miettiväiseksi. Hän haroo sormillaan

muutaman kerran punaista partaansa. Hän ei näytä niin
taipumattomalta kuin ennen. Tyttären kohtalo rentun vaimona
koskee sentään pohjimmaltaan hänen isälliseen sydämeensä. Ja
kun Fanny miehineen iltapuoleen palaa kaupunkiin, on hän saanut
sen puolen asiastaan toimitetuksi. Hänellä on viisi sadanmarkan
seteliä taskussaan ja loput on hän saava ensi viikolla, kaikki
äidinperintöjen nimellä. Hambergilla ja Rosinalla oli ollut keskinäinen
testamentti, joten jälkeenjäävä oli hallitseva jakamatonta pesää
kuolemaansa saakka, niinkuin Hamberg toisti kalunkirjoittajan
sanoja. Tämä oli sulaa hyvyyttä hänen puoleltaan.

Siihen toiseen asiaan oli Fannyn ollut vaikeampi päästä käsiksi.

Ukko oli suopea Fransiljalle, mutta eihän siitä vielä voinut ääntä
nostaa. Jannen Marilta hän uteli lähempiä tietoja ja Mari luetteli
kateellisena kaikki mitä Fransilja oli saanut lahjaksi sitten
markkinoiden, mutta sen enempää ei hänkään tiennyt. Mielellään
hän olisi juoninut, sillä Fanny oli luvannut hänen ensi lapselleen
vanhan kehdon, joka ajelehti aitan ylisillä muun joutavan rojun
kanssa ja Fransilja ei enää viime aikoina ollut niin kärkäs
kiehuttamaan kahvipannua hänen pistäytyessään siellä. Jaa —
kerran oli ukko sanonut Fransiljasta Kalle Krokströmille, joka oli
käynyt niitä kaaripuita tuomassa »että sen tytön ei tarvitse kauan
sulhasia odottaa», johon Krokström oli silmää iskien vastannut että
»mitäs niitä enää odottelisi kun on jo varma tiedossa». Tämä
tapahtui venerannassa, kun Krokström teki lähtöä; hän, Mari, oli ollut
siinä vieressä astioita pesemässä.

Ennen lähtöään pistäysi Fanny isänsä huomaamatta navetassa,

jossa Fransilja oli lypsyllä. Tyttö vavahti, mutta loi sitten
uhkamielisenä silmänsä rastaanpesään.

— »Sen minä sanon Fransiljalle että Fransiljan ei pidä panna

mitään hullutuksia ukon päähän, muuten Fransiljan käy huonosti.
Eikä Fransiljan pidä luulotella mitään isän rahoista, niiden päällä on
paperit ja lailliset testamentit. Että Fransilja tietää tämän etukäteen
eikä koetakaan mitään semmoista temppua. Muuten isä on jo yli-
ikäinen ja voi helposti joutua holhouksen alle, jos me omaiset niin
tahdomme. Ymmärtääkö Fransilja? Ja tietääkö Fransilja ettei
tämmöisessä naimisessa vaimo voi periä, ellei ole lasta ja mitä
siihen tulee — ääni kävi myrkylliseksi ja voitonvarmaksi — on se
mahdotonta. Isä on yli-ikäinen ja hänellä olisi ollut äidistä useampia
lapsia, jos… Se on lain mukaan tämä! Niin että Fransilja on nyt
järkevä ja katsoo eteensä niin ei minullakaan ole mitään Fransiljaa
vastaan.» Ääni koetti lopussa tavoitella joitakin lempeämpiä sointuja,
vaikkeivät ne ottaneet tullakseen.

Sanatulvan aikana oli tyttö suurilla silmillään tuijottanut Fannyyn.

Sydän jyskytti kuuluvasti röijyn alla ja korvia kuumotti niin ettei hän
ymmärtänyt puoliakaan tämän puheista. Lopuksi alkoivat huulet
vavahdella ja itkunsekaisella äänellä tuli:

— Minkätähden te kiusaatte minua!

Ja nyyhkytysten välistä:
 — Setä on itse kutsunut minut tänne. Mutta minä tahdon pois —
yhyy — jo huomispäivänä — yhyy — kun minua kohdellaan niinkuin
pahantekijää…

Ja jättäen kiulunsa siihen hän hyökkäsi Fannyn ohi juoksujalkaa

tupaan aikoen ilmoittaa lähdöstään Hambergille. Tupa oli tyhjä,
Hamberg oli mennyt jäälle sumppuavantoa luomaan ja silloin
Fransilja vaipui sängylleen ja itki siinä koti-ikäväänsä ja ihmisten
pahuutta, tyyntyen vasta vähitellen kun kissa tuli kehräämään hänen
olkapäätään vasten. Minkätähden hän koskaan olikaan lähtenyt
kotiseudultaan vieraiden ihmisten joukkoon! Minkätähden hän ei ollut
paremmin katsonut eteensä ja ajatellut mitä kaikkea tästä seuraisi.
Hamberg oli ollut ystävällinen hänelle ja hän oli ollut siitä hyvillään
ymmärtämättä mitään vaaraa. Oh, mitä kaikkea se siellä navetassa
puhuikaan! Nyt täytyi tästä tulla muutos!

Mutta samalla hänestä alkoi tuntua niin yksinäiseltä ja autiolta.

Mieli oli kuin viluinen meri kaukana jäänreunan takana.

Kun Patu vähän myöhemmin työnsi käpälällään oven auki ja

Hamberg sen jäljissä tuli tupaan oli siellä pimeätä ja hiljaista.
Hamberg sytytti lampun luullen Fransiljan olevan navetassa. Mutta
silloin huomasi hän tytön suullaan sängyssä.

— Fransilja, mikä sinun on?

Kun hän ei saanut vastausta, meni hän lähemmäksi ja kosketti

tytön olkapäätä. Ja kun Fransilja ei sittekään liikahtanut, alkoi hän
aavistaa asian laitaa. Ja hän pyyhkäsi tytön tukkaa niin pehmeästi
kuin taisi suurella karkealla kämmenellään ja sanoi eri äänellä kuin
äsken:
 — Fransilja!

Nyt nosti tyttö päätään ja katsahti häneen oudostellen.

— Mitäs me heistä, Fransilja!

Ja kun toinen itkua ennustavasti voihkasi:

— Minä tahdon pois täältä! tarttui hän Fransiljan käteen niinkuin ei

koskaan ennen ollut tehnyt ja käski kertoa kaikki, mitä he olivat
puhuneet. Siitä Fransiljan mieli vähitellen parani ja arastellen hän
kertoi Fannyn uhkauksista.

— Vai holhoukseen!

Hamberg ojentui suoraksi ja löi nyrkkinsä pöydänkulmaan.

— Ehkä Emilin hoitolapseksi!

Nyt hän jo nauroi, riisui saappaat jalastaan ja ripusti sukat

pellinnokkaan.

— Koettakoon heittiöt! ja siitä me pian saamme selon olenko minä

isäntä täällä ja vapaa tekemään mitä tahdon!… Mutta panes,
Fransilja, perunat tulelle. Rupesi niin hiukasemaan tuolla avantoa
hakatessa, hän lisäsi ripustaessaan takkiansa naulaan!

III.

Joulun alla talven aikeet vasta vakaantuivat… pakkanen kiristyi,

lunta tuprutti taivaan täydeltä yhteen menoon pari vuorokautta. Kun
Hamberg varhain kolmantena aamuna avasi tuvan oven oli koko
maailma ummessa. Navettatie ja kaivopolku olivat hävinneet, puut
seisoivat mäellä lumihupussa. Pienillä näreillä oli päässään vain
piippolakki, mutta lakkapäisillä petäjillä kokonainen kupooli.

Patu, joka oli maannut tuvassa pyrynajan, haisteli ja vainusi ovella

joka suuntaan. Sitten se iloisesti haukahdellen lähti uimaan
pehmeään valkoiseen hankeen niin että vain hiukan selkää ja
hännännipukka pisti lumesta. Väliin se loikkasi tasakäpälää
päästäkseen pikemmin, mutta silloin menivät silmätkin peittoon. Itse
tupa oli muuttunut lumipatteriksi, jonka harjalla peltinen kukko piti
vahtia, kääntyneenä päin tuulta ja kiekuen ruostuneella äänellään.
Savutötterö sen alapuolella huurusi hiljoilleen selkeään aamuilmaan
osoittaen että patterin sisässä kyti jotain elämää.

Hamberg käänsi silmänsä lumiselle lakeudelle, jota levisi

silmänkantaman kahteen suuntaan ja jota ei vielä yksikään
suksenlatu tai jalanjälki ollut rikkonut. Siellä seisoivat valkoisissaan
Aspöt ja Skytteskärit ja majakkaluodot, joihin saakka vain kettu oli
näihin asti uskaltanut kalaretkillään, sovittaen käpälänsijansa
suoraksi viivaksi härmäisellä jäällä.

Kun Hamberg oli lapiolla aukaissut tiet, meni hän katokseen, jossa
uusi vene oli alulle pantuna. Mielihyvissään silmäili hän vankkaa
emäpuuta ja kaaria, jotka vielä harottivat paljaina ilmassa kuin
lihattomat kylkiluut jossain suuressa kalanraadossa. Tämä
veneenteko oli ollut kauan haudottu aie, josta lopultakin tuli tosi.
»Kyllä se vanha Maija sinun ikäsi kestää» oli Kalle Krokström
sanonut, kun hän puhui tälle kaaripuiden tuomisesta. Mutta Hamberg
oli vastannut ettei hän halunnut viimeisenä saapua kalarantaan kun
se vanha lata oli niin hapan ja vettynyt, että kääntyi kankeasti kuin
santajaala eikä noussut enää tuuleen niinkuin piti. »Vielä minä yhden
veneen kulutan iässäni», hän oli sanonut ja tosi siitä oli tullut.
 Tänään hän tahtoi oikaista selkänsä sisätöiden jälkeen. Hänen
lyödessään ensimäisen naulan kaitaan, vastasi iloinen kaiku
rannoista ja salmen takaa. Patu haukkui jotakin jäällä, lähimmällä
oksalla käänteli tiainen päätään ja äänteli yksikantaan tint-tint.
Lehmä ammahti kerran navetassa; vajasta kuului Jannen puukirves;
molemmista tuvista nousi savu sinertävänä ja suorana ilmaan.

Hänessä on koko aamun, koko viikonkin ollut hiljainen hyvämieli.

Passatessaan, tähdätessään, naulatessaan hän suunnitteli —
passasi ja
käänteli päässään ajatuksia aivankuin lautoja ja kaluja käsissään.
»Noinpäin se menee, ei paremmin sittekin näin.»

— Siinäpäs kerran kuulivat Fanny ja Emil! Suoraan ja selvään oli

herrassöötinki heille laukassut!

Asia oli niin että Hamberg oli, heti tyttärensä ja vävynsä käynnin
jälkeen viime viikolla, mennyt vanhan tuttavansa, herrassöötinkin
puheille, joka kesäisin asui huvilassaan naapurisaarella ja jonka
kanssa he monasti syksyisin olivat olleet allin kuvilla. Herrassöötinki
oli suopeasti kuunnellut hänen asiansa, oli sanonut ettei kukaan voi
estää häntä naimasta ja käyttämästä itse ansaitsemiaan rahoja ja
luvannut Hambergin pyynnöstä esittää lain puustavin tyttärelle ja
vävylle.

Se päivä oli Hambergin mielihyvän salainen lähde. Hän oli

kuvannut moneen kertaan kohtauksen Fransiljalle.

He olivat, Fanny, Emil ja hän istuneet kolmisin kauniissa salissa,

suuren kivimuurin toisessa kerroksessa. Siellä oli ollut paljo
katseltavaa: kukkia, tauluja, kuvapatsaita. Emil ja Fanny olivat olleet
kuin hiukan hämillään, tervehtineet lyhyeen, supisseet vain pöydän
luona keskenään. Hän, Hamberg, oli istunut erillään heistä
kiiltonappisessa takissaan, lakki vieressään lattialla. Silloin oli
herrassöötinki ankarannäköisenä tullut sisään, kädessään
Hambergin testamentti ja kalunkirjoitus, katsahtanut pulleilla
silmillään paperien yli Fannyyn ja Emiliin ja sanonut: »Kuulkaapas,
rouva ja herra Karlsson, teidän tulee elää sovussa isä Hambergin
kanssa, sillä hänellä on semmoiset paperit että hän voi lahjoittaa
vaikka ventovieraalle viimeisen kolikkonsa ja silloin käy niin ettette
saa tuon enempää — tässä herrassöötinki puhalsi pitkin
kämmentään. Ja mitä siihen toiseen asiaan tulee, sanotaan sanassa
selvästi: lisääntykää ja täyttäkää maa!» Mutta tätä sanoessaan oli
herrassöötinki ollut vielä tuikeamman näköinen, silmät olivat olleet
pulpahtaa päästä ja niska oli paistanut leveänä ja rasvaisena kun
hän käveli edestakaisin lattialla. Fanny ja Emil olivat istuneet
tuoleillaan, punaisina kuin saunanlöylyssä ja hyvästiä sanomatta
olivat he sieltä lähteneet.

Hamberg on elänyt kohtauksen uudelleen, hän on hymähtänyt

itsekseen muistaessaan miten höyli ja hyvätuulinen herrassöötinki oli
ollut sitten jälestäpäin tarjotessaan hänelle sikaarin ja kahvit ja
laskiessaan leikkiä Fransiljasta. Oikea rehti mies ja tottunut meri-
asioissa se herrassöötinki! Niillä linturetkilläkin sillä oli semmoisia
pieniä konjakkipulloja taskuissaan.

— Kas pahusta!

Vasara oli luiskahtanut ja naula mennyt vääräksi. Hän kiskasee

sen reiästä ja pistää hampaiden väliin vääntääkseen suoraksi, kuten
tavallista, mutta tällä kertaa on onni kova. Pari hammasta ylä- ja
samanverran alaleuasta karahtaa poikki. Hamberg sylkäsee palat
suustaan käsittämättä oikein mitä on tapahtunut, ennenkuin tuntee
hiljaista mojotusta, joka yltyy raivoisaksi repimiseksi. Hän kourasee
suutaan.

— Kas perkelettä, kun vei parhaimmat leipähampaani!

Ja noituen »valssattuja» rautanauloja hän menee tupaan

pyytämään Fransiljalta hammastippoja. Mutta kolotus vain yltyy. On
kuin joku tahtoisi vääntää leukaluun sijoilta tai porata sen
kappaleiksi. Tulisia kipinöitä alkaa tanssia silmissä, ei kärsi enää
istua yhdessä kohden, vaan täytyy kulkea ympäri tupaa suuri
villahuivi pään ympäri käärittynä, huojutella päätään ja ähkiä.

Mutta kun hän ei saa istua eikä maata eikä muutenkaan olluksi,
juoksee hän ulos pakkaskirkkaaseen yöhön ja sieltä kala-aittaan kuin
hädissään hakien jotakin. Oven suussa on silakkanelikko. Siitä hän
sieppaa pari jäätynyttä kalaa ja purra rouskahuttaa ne hampaiden
välissä ja on kuin kolotus helpottaisi tästä oudosta lääkkeestä.

Se oli Hambergin ensimäinen hammastauti, jota hän kauhulla

muisteli kauan jälkeenpäin ja joka hänen muuten ehjään
hammasriviinsä jätti ammottavan sylkyaukon.

*****

Tuuli oli jouluyönä käynyt lännestä, joten saattoi toivoa hyvää

kalavuotta. Hiljaiset joulupyhät oli kalastajatuvissa maattu ja syöty,
uutena vuotena oli otettu humalat, samoin loppiaisena, jolloin nuoret
olivat tanssineet hanurin mukaan suurimmissa tuvissa. Sitten tulivat
härkäviikot pitkine puhteineen ja verkkotöineen.

Eräänä iltapuolena olivat Kalle Krokström tyttärineen ja Hambergin

veli Kustaa poikenneet Tallskäriin, jossa pikku-Vinter oli pareillaan
syöttejä kaupalla. Pikku-Vinter oli entinen alaupseeri, joka oli
koetellut maailmassa yhtä ja toista ja nyt ikäloppuna elätteli itseään
syöttikalastuksella ja koukkujen teolla. Istuttiin pöydän luona,
savuavan kattolampun alla. Patu loikoi keskellä lattiaa, missä kissa
leikitteli sen hännänpäällä. Koira oli olevinaan äkäinen, nousi joskus
etukäpälilleen näyttämään punaisen kielensä, mutta silloin kissa oli
tähtäilevinään katossa olevaan kalakassiin, johon Hamberg oli
Fransiljan iloksi pyytänyt tilhen ulkoa pihlajasta. Tilhi istui niska
kyyryssä orrellaan, vihelsi väliin haikeasti eikä koskenutkaan
vieressä olevaan pihlajaterttuun.

— Sen nyt tiesi ettei Anselmillekaan hyvin kävisi! sanoi Hamberg

salaperäisen merkitsevästi. Nyt on jo kolme pojista kuollut ja viime
syksynä tuli tytär Anna hulluksi. Bengt taittoi jalkansa iljangolla,
Eerikki lähti merille ja jäi sille tielleen! Ja ilmankos olisi Storössä niin
paljo kummitellut aikoinaan!

Kalle Krokström ja Kustaa nyökyttivät päätään, he olivat kumpikin

senaikuisia miehiä ja tunsivat asian. Mutta Fransilja ja Lyydi
kuuntelivat höröllä korvin ja kysyivät yhteen ääneen, mikä siellä
Storössä sitten oli kummitellut.

Kuka sen tarkkaan tiesi! Toiset sanoivat että se oli reppuryssä,

jonka Korander yhdessä vaimonsa kanssa oli tappanut ja haudannut
perunakuoppaan, toiset että se oli eräs Kirkkonummen
kalastajaukko, jonka Storöläiset olivat yöllä ottaneet hengiltä
saadakseen hänen sumppunsa joka oli täynnä haukia. Koranderin
itsensä kävi huonosti, hän hukkui heikkoihin jäihin ja vaimo Anna ei
kolmeen vuoteen ummistanut silmiään öillä, vaan väänteli itseään
sängyssä voivotellen: »Voi voi voi, koko mun joukkoni, koko mun
joukkoni palavaan helvettiin, palavaan helvettiin!» Tätä hän huuteli
yöt läpeensä katsoen jäykillä silmillään ja kuulematta mitään, jos
joku häntä puhutteli. Päivisin hän oli päästään selvä niinkuin muutkin
ihmiset. Että siinä talossa eivät asiat olleet oikein tiesi siitäkin että
rukki alkoi yöllä itsestään pyöriä, vesiämpäri lähti liikkeelle
nurkastaan ja seinän takaa kuului naputuksia. Storö oli ennen ollut
oikea yöpaikka pitkämatkaisille, mutta kun tämmöistä alkoi kuulua, ei
sinne poikennut enää kukaan, vaan sousi tavallista kiiruummin läpi
salmen, jonka rannalla oli musta sauna.

Siinä saunassa oli hän itse, Hamberg, nähnyt pahan pitävän

peliään. Hän oli isänsä kanssa ollut tuulastamassa Storön lahdella
syyskuun yönä, kun ukko oli huomannut tulen loimottavan saunan
akkunasta. »Mitähän nyt Koranderilla on tekeillä, kun on tuli
saunassa yölläkin», oli ukko sanonut ja he olivat huovanneet rantaan
katsomaan. Mutta sauna olikin tyhjä ja kiuas kylmillään. Vaan tuskin
olivat he päässeet rannasta, kun loimotus taas näkyi ja koko ajan
minkä he liikkuivat lahdella aina kotirantaan saakka. Tiesi sen ettei
semmoista turhaan tapahtunut! Jotain viatonta verta oli Storöläisten
tunnolla ollut!

Nyt puuttui puheeseen pikku-Vinter, joka kaupungista oli tuonut

tiedon Anselm Koranderin tapaturmaisesta kuolemasta. Hän arveli
että jos ryöstetty tavara olisi toimitettu takaisin surmatun omaisille ja
kolmena pyhänä peräkkäin käyty Herran ehtoollisella, olisivat
kummittelut lakanneet ja Storöläiset päässeet vähemmillä
onnettomuuksilla. Itse hän muisti tapauksen Skoglannista, tiilitehtaan
ajoilta, miten kuollut oli tiennyt omaansa kaivata. Siellä oli ollut mies,
Reuhka-Kalle nimeltään, rokonarpinen ja rietas suustaan. Hän oli
tiilikiertoa ajamassa ja sama Reuhka-Kalle oli rautakangella
katkaissut kapteeni Brummertin valkoisen tamman selän, kun se ei
totellut hänen hoputustaan. Nimensä oli hän saanut siitä että kesät
talvet piti päässään vanhaa karvareuhkaa, ja muuta omaisuutta ei
miehellä juuri ollutkaan. No, kerran sitten olivat tiilentekijät pyhäiltana
juopottelemassa Pannukakulla, siinä Skoglannin luona, kun jostain
syntyi riita ja Reuhka-Kalle lähti vaatteissaan uimaan pakoon
Skoglannin puolelle. Toverit tulivat kuitenkin perässä veneellä ja
hakkasivat airolla päähän niin että Kalle heitti henkensä ja upposi
salmeen. Hän, Vinter, oli ollut onkimassa ylös ruumista, joka pantiin
ranta-aittaan, siksi kunnes Öhman ehtisi tehdä kirstun. Mutta
sielläkös rupesi kummittelemaan ja kuulumaan outoja ääniä!
Brummertin piika oli nähnyt kuinka aitan harjalla oli istunut
kokonainen rivi pikkupiruja ja aamulla kun mentiin katsomaan oli
ruumis kääntynyt laudoillaan. Tiilentekijät tuumivat keskenään että
mikähän sitä Kallea vaivaa ja mitähän se kaipaa, kunnes joku keksi
että se kaipaa reuhkaansa, kun se ei eläissäkään raaskinnut ottaa
sitä päästään! Ja niinpä olikin: kun reuhka haettiin kuolleen päähän,
sai ruumis rauhan eikä aitassa enää kummitellut, lopetti Vinter
kertomuksensa.

Nyt huusi Kalle Krokström, joka tunsi käryä nenässään, että puuro
palaa pohjaan. Fransilja riensi punehtuen lieden luo ja tarttui
kauhaan. Miehet panivat uutta tupakkaa piippuun näyttäen tavallista
miettiväisemmiltä.

— Eipä luulisi että syntinen ihminen enää pääsee sieltä

kummittelemaan, katkaisi Kalle Krokström vaitiolon, mutta on sitä
kuultu niin monta kertaa että uskoa täytyy! Aspön Lindholm-vainaja
kertoi minun nuoruudessani että hänellä oli veli, jonka kanssa olivat
kerran tehneet semmoisen välipuheen että se joka ensinnä kuolisi,
tulisi toiselle kertomaan olostaan haudan takana. Veli oli kippari, ajoi
santajaalansa karille ja upposi eräänä marraskuun yönä. No, kului
muutama vuosi ja Lindholm oli unohtanut asian, kun eräänä talvi-
iltana tämmöisenä kuin tämä, pistäytyy paitasillaan pihalla — millä lie
ollut tarpeillaan — ja velivainaa oli tullut hiljaa kuin hiihtäen ja aivan
valkoisena nurkan takaa sanomaan: Vieläkös Petter tunnet minut!
Mutta Lindholm-vainaalta oli luonto mennyt niin tyystin ettei saanut
ääntä kurkustaan ja polvet olivat ruvenneet tutisemaan. Oli kuin tulen
alta juossut tupaan, vaan veli oli tullut perässä ja pidellyt ulkoapäin
ovenripaa. Silloin oli jättänyt oven raolleen ja hypännyt suinpäin
sänkyyn akan selän taakse. Ja vasta pitkän ajan kuluttua, kun ei
ketään kuulunut oli peitteen alta kurkistanut, tönässyt akan hereille ja
sanonut: Vedäs tuo ovi kiinni, koirapakana kun juoksi sen auki!
Haamusta ei näet uskaltanut hiiskua sanaakaan.

Fransilja on koko illan kuunnellut miesten juttuja. Joskus kulkee

kuin kylmänväreet hänen selkäpiitään pitkin, ja kun vieraat ovat
lähteneet ja illallinen syöty, alkaa häntä pelottaa tuvan hiljaisuus.

Uni ei tänään ota tullakseen. Ulkona on suojatuuli, vesi tippuu

räystäältä, uuninpelti kolahtaa silloin tällöin. On kuin joku huokaisi
seinän takana… ehkä se kuitenkin on jää joka ääntelee veden
noustessa. Ei nyt kuuluu selvään huutoa… Fransilja katsoo silmät
kauhusta jäykkänä kuinka ikkunan hiukan häämöttävästä neliöstä
tulee valkoinen haamu, silmät kuopalla. Hän parahtaa ja juoksee
kamariin. Hamberg herää siihen että joku ryömii hänen peitteensä
alle ja painautuu kiinni hänen kylkeensä. »Minä en uskalla olla yksin
tuvassa», saa Fransilja hänelle kuiskatuksi.

Sinä yönä nukkuivat Hamberg ja Fransilja ensi kertaa yhdessä.

*****

Kuulutuksiinpano viipyi ensi avoveteen, sillä Fransilja oli ennen

ratkaisevaa askelta tahtonut kuulla vanhempiensakin mieltä asiassa