6.346 Rev: The Parameter of Aspect (C. Smith)

[The Linguist List]

----------------------------------------------------------------------
LINGUIST List:  Vol-6-346. Thu 09 Mar 1995. ISSN: 1068-4875. Lines: 300
 
Subject: 6.346 Rev: The Parameter of Aspect (C. Smith)
 
Moderators: Anthony Rodrigues Aristar: Texas A&M U. <aristar at tam2000.tamu.edu>
            Helen Dry: Eastern Michigan U. <hdry at emunix.emich.edu>
 
Review Editor: T. Daniel Seely: Eastern Michigan U.<eng_seely at emunix.emich.edu>
 
                            REVIEW EDITOR'S NOTE:
 
What follows is another discussion note contributed to our Book Discussion
Forum.  We expect these discussions to be informal and interactive; and
the author of the book discussed is cordially invited to join in.
 
If you are interested in leading a book discussion, look for books
announced on LINGUIST as "available for discussion."  (This means that
the publisher has sent us a review copy.)  Then contact Daniel Seely at
     eng_seely at emunix.emich.edu
 
-------------------------Review-------------------------------------
 
The Parameter of Aspect (POA) Carlota S. Smith    Kluwer 465p
Reviewed by Robin Schafer
 
Overview
     This book is divided into two parts. In the first, an
approach to the treatment of aspect is presented; in the second,
this approach is applied to five languages --- English, French,
Russian, Mandarin, and Navajo.
     The discussion in Part One, while mentioning all five of
these languages, is illustrated primarily with English examples.
Here a "Two Component Theory" of aspect is presented and formalized
in Discourse Representation Theory (DRT). The temporal
organization of a situation is discussed with respect to three
factors: "situation type", "aspectual viewpoint", and a
subjective set of factors discussed briefly, but essentially
outside of the account presented. Situation type and aspectual
viewpoint are the two components of the theory presented.
Situation type refers to the type of event or state involved
in an utterance. The temporal perspective presented by the
speaker  --- completive, incompletive, or neutral --- is what
is referred to as aspectual viewpoint.
     The core proposal is that aspectual meaning results from the
interaction of these two independent components. The approach
is presented in overview in Chapter One, and more detailed
background and explanation of the two components are provided
in Chapters Two through Five. Part One closes with a
formalization of the approach in Chapters Six and Seven. Part
Two consists of five chapters, each focusing on the aspectual
system of one of the five languages listed above. I overview
primarily Part One here.
     Situation types include the classic state, activity,
accomplishment and achievement, plus the category semelfactive
(instantaneous atelic events exemplified by `knock' or `cough').
These are defined through use of conceptual features each with a
plus or minus feature value: [stative], [telic], [durative]
following current work. Original to the POA account is the use
of what is called the "basic-level category of V" to define
central cases of a situation type. Basic-level categorization
need not be adhered to in actual sentences; it is the simplest
and most direct association of verb arguments with an idealized
situation type. The association is often based on stereotypes or
conventional conceptions of events. The notion of basic-level
category is consistent with the assumption that aspectual categories
are based on human cognitive abilities rather than being language
dependent.
     Chapters Two and Three contain lucid discussion of all five
situation types, distinguishing type from both situations themselves,
and the sentences that express them. The subsection relating the
cognitive notion of clausal chain to basic-level categorizing is
particularly helpful in clarifying the perspective taken on these
notions. I would have also appreciated some brief comment on what (if
anything) was the current consensus in the field concerning these
notions and their representation, and how the work in POA fit with it.
     Situation type information is made "visible" by viewpoint aspect.
Types of viewpoints differ in terms of how much of a situation they
make visible. A significant portion of Chapter Four is devoted
to distinguishing three viewpoints, the perfective, the imperfective,
and a newly proposed "neutral" viewpoint.
     Sentences that have neither perfective nor imperfective marking,
and which are often referred to as aspectually vague, are here
presented as having a default neutral viewpoint. Given the system
presented in POA, this is theoretically appealing: every sentence
must be provided with an aspectual viewpoint since situation type
information is made available only through viewpoint. Empirical
support for this third category involves the range of meanings
found in aspectually vague sentences: they allow both open
(imperfective) and closed (perfective) readings. Examples are given
from French and Chinese (and Navajo in Chapter Twelve). I repeat the
French example. In when-clauses the future tense is used with
situations that are interpreted as either open or closed:
 
      Jean chantera quand Marie entrera dans le bureau.
      Jean singFUT  when  Marie enterFUT in  the office
     `Jean will begin singing when Marie enters the office.'
     `Jean will be singing when Marie enters the office.'
 
However neutral viewpoints are also shown to be less than entirely
flexible: for example, they do not focus on the preliminary stages
of a situation, a reading often associated with imperfective
viewpoint. This type of data is crucial to the argument, for without
it the sentences can simply be said to have no viewpoint and are
freely interpreted. Given these data, it is further suggested that
the endpoints of a situation type (preliminary stages, end results,
etc.) are not visible to the interpretation of sentences with neutral
viewpoint.
     The contribution of pragmatics to the interpretation of
viewpoints is nicely discussed in a concise fashion in the final
section of Chapter Four. This is discussed in further detail in
Chapter Ten, co-authored with Gilbert Rappaport, on the Russian
aspectual system.
     Chapter Five examines the interaction of aspectual meaning and
temporal meaning introduced by tense marking and adverbials. In some
respects, these are complementary temporal systems since the temporal
organization that aspect is concerned with, unlike tense, involves the
internal structure of a situation. Viewpoint information may also be
conveyed by tense, and in some languages tense contributes directly to
aspectual values. Chapter Nine on the aspectual system of French,
focuses on these interactions of tense and aspect. Even indirectly,
the interaction can be extensive, for example, perfect tense is
presented as involving both temporal location and aspectual
information in that it derives a stative situation type.
     Situation type and aspectual viewpoint are informally
represented in Chapters Two through Five (and in the discussion in
Part Two) using a "temporal schemata". Distinguishing features
characterizing situation type are represented as different points
in the unfolding of a situation. For example, the initial and final
points are indicated by I and F respectively, the internal structure
of an event is represented with dots joining I and F, in statives I
and F are joined by an underscore, a result is represented by R, lack
of duration is represented by placing I over F. Viewpoints are
indicated by well-placed slashes. Thus we find:
 
state         (I)_____(F)activity       I.....F(arb)
accomplishment I.....F(nat)(R)
semelfactive   I
               F
achievement    ...I (R)...
                  F
perfective     ///////
imperfective   ...////...
 
So: the example `Mary walked to school.' is represented  I....F(nat)R
                                                         ///////////
 
the resultative reading (it's already seated) of `The cat was sitting
in the chair.' is represented: I...F../////...
     These schema serve an expository purpose but also foreshadow the
formalizing of aspectual concepts as temporal intervals in Chapter
Six. I found the discussion of the relation between the temporal
schemata and temporal intervals, and temporal intervals and DRTs less
than clear. The final point in formalizing a representation of
aspectual information is to integrate its representation into some
larger semantic representation -- here, the discourse representation
structure (DRS). Events and states are introduced into the DRS as
entities corresponding to the five situation types. That is, they are
characterized as states, activities, etc. This characterization occurs
according to the (intensional) temporal properties ascribed to the
entity. It is the meaning of these properties that are analysed as
temporal intervals in Chapter Six.
     Situations unfold in time, and their properties are manifested
as this takes place. A general set of intervals with which the
temporal development of a situation type can be expressed is assumed.
These included at least,
 
i-1         in which the situation does not obtain,
i           the initial point of the situation,
i+1,2,etc   the internal stages,
i+1...n     the duration (without internal structure)
j           the final point
n           the natural final point
j+1         in which the situation does not obtain,
n+1         in which the resultant state obtains
n+2         in which the situation does not obtain
 
Each of the five situation types can be expressed as a combination of
certain of these intervals. For example, an accomplishment includes
i-1; i; i+1,2, etc; n; n+1. Viewpoints are located at independently
specified intervals. For example, the imperfective is located at an
interval such that all the times in that interval are times when an
interval of a situation obtains, and there are no times at which the
endpoints of the situation obtain. (So imperfective viewpoint does not
include i-1; j+1; or n+2.)
     What this formalism does is more explicitly define the unfolding
of a situation in time. The interaction of the two components modeled
here is essentially the same as that modeled by the temporal schemata.
As far as I understand, the stipulation that "situation types in
sentences require a viewpoint to make them semantically visible"
remains a stipulation. Also note that viewpoints must preserve
information about situation type. Thus it is stipulated that
the interval defined by viewpoint inherits the intensional situation
type properties of the situation entity. (This is at least conceptually
made easier by the interval analysis of those properties.) After
Chapter Six, however, the temporal interval analysis of the properties
characterizing situation type is assumed in the background, and plays
no direct role.
     Determining aspectual meaning involves at least: (i) intrinsic
aspectual features assigned in the lexicon; (ii) a syntactic
representation; (iii) compositional rules constructing semantic
representation from the syntactic representations; (iv) association of
these representations with meanings; and (v) integration into some
larger semantic representation. The second half of Chapter Six
discusses (i) through (iv), with a particular emphasis on
composition rules. The compositional rules compute the  value of
situation type and viewpoint from the input of (i) and (ii).
(Viewpoint features are directly associated with meanings; the meaning
corresponding to a situation type is associated with a predicate
phrase (more specifically: a "verb constellation").)
     The  algorithm for integrating this analysis into the DRT
framework is presented in Chapter Seven. The choice of DRT is said to
be primarily due to the fit in conceptual representation (DRT
develops both a truth conditional interpretation and a conceptual
representation). While the viewpoint information is introduced
as an interval into the DRS, events and states are introduced as
entities corresponding to the five situation types. Other temporal
concepts are also incorporated into example DRSs, demonstrating the
approach to the interaction there.
 
Useability
     Confusion between aspectual classes and aspect (viewpoint)
morphemes as sources of aspectual information frequently arises in
discussions of verbal morphology and morpho-syntactic features. A very
appealing characteristic of this book is simply the treatment of both
aspectual classes and aspectual viewpoint, and the clarity with which
these are both shown to reflect temporal organization. As such it is a
valuable reference. Further, its organization is well-conceived: the
separation of the discussion of the phenomenon and its analysis from
applications of the analysis to specific language data makes it easy
to access specific information. It also allows for in depth discussion
of non-English language facts without interruption of general or
background discussion.
     The time spent on introduction and explanation of the fundamental
notions of aspectual classes of verbs and viewpoints, plus the clear
and concise introduction to DRT in Chapter Seven enhances use of the
volume as a text. The data and its discussion in Part Two further adds
to this appeal, as well as to the use of the book as a reference.
     Nonetheless, the work is marred by poor editing, much of which
should have fallen within the purview of a copy editor. The number of
"typos" is appalling. There are the disruptive but otherwise harmless
kind, for example words remaining in sentences that have clearly been
rewritten, misspellings (often final letters of a word are missing),
agreement mismatches, extra or missing spaces, remaining portions of
typesetting commands (e.g. "dots" following a name requiring an
umlaut), botched punctuation (ellipses seem to have been a
particular challenge to the typesetter), page numbers missing from the
table of contents (!), and inconsistency in the presentation
(typeface, etc.) of non-numbered sub-sections. These should all have been
caught by a competent copy editor. More serious editing problems
involve mistaken figure numbers in text references, footnote numbers
missing from the text, names missing from the name index, (names
misspelled in the name index), a large number of incomplete or
erroneous entries in the general index (including entries under
features such as detachability, non-detachability, and atelic), a few
missing entries in the general index (including acquisition, negation,
and temporal schema(ta)), incorrect line numbers in reference to DRSs
(reference to line 11 and 12 of a DRS with 10 lines). These are
embarrassing mistakes, and particularly those to do with the indices,
lessen the overall useablity of the volume. (I only performed spot
checks of indices and references. I found no missing references.)
     However, in my opinion, the most serious presentation problems
are three. There are unexplained variations in the informal temporal
schemas. This is significant given that these are heavily relied on
for expository purposes throughout the book. Second, there is an
unnecessary proliferation of terminology in certain discussions. (For
example, following the presentation of the feature [+/- stative] in
the text, it is included in a chart on the next page as the feature
[+/- static], after which [- static] is referred to as [+ stage].)
Finally, non-English examples --- including those is Part Two of the
book --- are rarely provided with both a morphological gloss AND a
translation. This is absolutely crucial to the use of the work for
reference purposes and also aids in the complete understanding of the
data.
 
Discussion Questions
     In addition to material presented in the Overview, particularly
the three innovative elements of the work, two questions warrant
general discussion. The first involves the title of the work. It
remains unclear to me why the term parameter appears, and how aspect,
as it is presented here, is conceived of as a parameter. In POA it is
suggested that the viewpoint component alone is parameterized, and
this, in a two-component system of aspect, would seem to call into
question the existence of a parameter of aspect. Secondly, among
issues concerning the interaction of syntax here, the assumption that
it is s-structure representations, rather than LF representations,
that are the input to compositional rules requires further
consideration.
 
--------------------------------------------------------------------------
LINGUIST List: Vol-6-346.