496 College & Research Libraries September 2010

of real-life assessment in practice. Instruc-
tion librarians and coordinators will find 
helpful assessment models to follow in 
this volume. The assessment tools de-
scribed in this book are not necessarily 
new and different; rather, what differ-
entiates this volume is how the assess-
ment tools are used in faculty-librarian 
partnerships, so this book will especially 
be informative and practical for librarians 
seeking to partner with teaching faculty to 
enhance and assess information literacy 
learning.—Maria T. Accardi, Indiana Uni-
versity Southeast.

Julian Warner. Human Information Re-
trieval. Cambridge, Mass.: MIT Press, 
2010. 189p. alk. paper, $35 (ISBN 
9780262013444). LC2009-010120.

In Human Information Retrieval, Julian 
Warner, a faculty member in the Man-
agement School of Queen’s University 
at Belfast with a long-standing interest 
in the intersection of humanistic and 
technological approaches to information 
systems, undertakes the ambitious project 
of rethinking the theoretical framework 
of information retrieval (IR). Judging 
traditional approaches to the topic as 
being overly limited in their concerns or 
based on theories insufficiently congru-
ent to real-world practice, he seeks to 
develop a comprehensive perspective 
on IR that both acknowledges its human 
dimensions and gives a theoretically 
adequate account of currently dominant 
retrieval practices. Over the course of nine 
chapters, Warner unfolds this agenda 
by (1) outlining a new approach to IR 
centered around the concept of human 
labor and (2) articulating a rationale for 
full-text retrieval, which, in his view, is the 
characteristic modality of IR deployed in 
information technologies today.

After an introductory chapter, in which 
he carefully outlines the trajectory of ar-
gument for the book as a whole, Warner 
devotes three chapters to an explication 
of his labor-theoretic approach to IR. In 
his view, enhancement of selection power, 
defined as “the human ability to make 

choices between objects or representa-
tions of objects,” is the primary aim and 
core value of IR. Selection power depends 
upon selection labor, the mental work that 
goes into the processes of discriminating 
between objects and choosing between 
them. Although selection labor is an activ-
ity of the human mind, some of its pro-
cesses can be supported by and, in part, 
transferred to information technologies. 
Such transfers are possible, in Warner’s 
estimation, because these technologies 
are themselves the products of human 
labor and designed with specifically hu-
man purposes in mind: historically, there 
has been a tendency to delegate as many 
processes of human labor as possible to 
technological tools.

Selection labor can be further subdivid-
ed into two mutually exclusive types: de-
scription labor and search labor. Descrip-
tion labor is the material and mental work 
expended in “transforming objects (docu-
ments, images, or people) into searchable 
descriptions that will assist subsequent 
retrieval,” whereas its converse, search la-
bor, is the material and mental work spent 
in searching for objects in an information 
system. In a library setting, for example, 
the cataloger preparing (or modifying) a 
catalog record is engaging in description 
labor, while the user posing a query to 
an online catalog or a search of engine is 
involved in search labor. 

Crosscutting the distinction between 
description labor and search labor is a 
further distinction between semantic and 
syntactic labor. Semantic labor is mental 
labor pertaining to the production and 
determination of meaning, while syntactic 
labor is mental labor used to carry out 
operations based on pattern matching 
and the manipulation of symbols. Both 
descriptive and search labor have seman-
tic and syntactic components: for instance, 
the assignment of an index term to a docu-
ment requires semantic description labor 
and the arrangement of a set of records 
into a series by the alphabetical order 
of their entry terms involves syntactic 
description labor, whereas translating a 



Book Reviews 497

question into an authorized query term 
represents a kind of semantic search 
labor and the sorting of a set of retrieved 
records by date is a type of syntactic 
search labor. According to Warner, a fun-
damental distinction between semantic 
and syntactic labor lies in the fact that the 
former is “irreducibly human” and not 
amenable to mechanization, whereas the 
latter is “transferable to technology as [a 
form of] machine process.” An important 
corollary of this view is that, ultimately, 
there are definite limits to the transfer of 
selection labor to information technology: 
in other words, human mental labor can-
not be entirely evaded in IR. 

After laying out the fundamentals 
of his new approach, Warner devotes 
the final five chapters to developing a 
theoretical account of full-text retrieval 
from within a labor-theoretic perspective. 
His point of departure is the difference 
between word and phrase searches on 
Internet search engines. Basing himself 
on a series of concrete examples, he ob-
serves that the use of single keywords in 
searches tends to yield sources on diverse 
topics, whereas the results of phrase 
searches manifest greater semantic homo-
geneity. Such results indicate that isolated 
keywords are multivalent in their mean-
ing, while phrase searching counteracts 
this semantic multivalence by inserting 
words into contexts that constrain their 
meaning. Arguing that this phenomenon 
is common to the practical experience of 
full-text search but has not been given suf-
ficient theoretical attention by librarians 
or information scientists, Warner draws 
upon Saussurean linguistics and informa-
tion theory to explain the semantic and 
syntactic bases of retrieval from full text. 

 Warner’s account of the semantics and 
syntactics of full-text retrieval is intricate, 
and limitations of space preclude any 
detailed exposition of it here: neverthe-
less, an indication of the main lines of 
his model of semantics can provide at 
least a taste of his approach. According 
to Warner, phrase searches deploy what 
Saussure termed the syntagma: that is, the 

linear string of words naturally produced 
in human speech and writing, in which 
the meaning of each word-element is 
determined by its relation to the other 
elements in the string. To remove a word 
from any syntagmatic context, as is done 
in single keyword searches, is to shift the 
basis of its meaning to what Saussure 
termed the paradigm, or the network of 
all possible meanings that a word might 
have within a given language. Entering 
into a paradigm releases a word from the 
semantic constraints of the syntagma, and 
it is this that accounts for the semantic 
heterogeneity of resources yielded by 
keyword searches. Full-text retrieval, 
then, involves varying degrees of connec-
tion between linguistic expression and se-
mantic content that can impact the results 
of search in different ways. Significantly, 
though, both types of searches require 
semantic search labor, for the onus is on 
the searcher to disambiguate and inter-
pret the results of a full-text search. This, 
Warner believes, stems from the fact that 
“human understanding and composition 
of written language may be highly intrac-
table computationally”—a thesis that fits 
neatly within the labor-theoretic view that 
semantic labor is the irreducibly human 
element of selection power. 

Of special interest to readers of this 
journal will be Warner’s view of the value 
of human-assigned metadata for IR, an 
issue that has stirred much controversy 
in the library community in recent years. 
In his view, traditional mechanisms for 
generating human-assigned metadata, 
such as catalogs and controlled vocabu-
laries, have several disadvantages: they 
are grounded in an (over-)simplified 
and deterministic view of language as 
nomenclature; their protocols and prod-
ucts are limited—and limiting—legacies 
of a premodern technological régime; and 
they involve semantic description labor, 
which is costly to support. Nevertheless, 
he acknowledges that the products of 
semantic description labor do have value 
in collocating materials that would oth-
erwise not be retrieved together; accord-



498 College & Research Libraries September 2010

ingly, he concludes that they may have 
a continuing, if diminished, role to play 
in an informational environment that in-
creasingly privileges systems for full-text 
retrieval such as Internet search engines. 
Because, in Warner ’s view, semantic 
labor is categorically not transferable 
to machines, any reduction in semantic 
description labor entails an increase in 
semantic search labor. Warner tends to 
view ongoing trends toward increased 
semantic search labor with equanimity: 
not all members of the library community 
will share this attitude.

All in all, Human Information Retrieval 
presents its readers with an interesting 
perspective on IR that well repays study. 
The emphasis on human labor as a central 
analytic category for understanding IR 
processes allows Warner to integrate a 
number of disparate themes that have 
long been objects of reflection for theorists 
of librarianship and information science 
into a single, fairly compact theoretical 
framework. Certainly, the labor-theoretic 
approach, as set out in the opening chap-
ters of the book, is structurally elegant 
and has the potential to be a very useful 
model for “macroscopic” thinking about 
the design and evaluation of IR systems. 
One senses, though, that it is not yet com-
plete and requires further development 
if it is to fulfill its author’s ambitions: for 
example, it pays scant attention to the 
different types of (and motivations for) 
search—a factor that should certainly be 
taken into account in any theory intended 
to support the evaluation and design of 
IR systems. 

Finally, it should be noted that the 
book is written in an idiom both erudite 
and abstract: for this reason, it is a de-
manding read. Laudably, Warner tries to 
ease the reader’s way through the book 
by providing concrete examples, culled 
from a variety of sources, to illustrate 
the theoretical points he is making, em-
ploying diagrams and pictures to aid in 
the visualization of certain points, and 
constantly signposting and summarizing 
the key arguments: moreover, in addition 

to a bibliography and a (rather indiffer-
ent) index, he helpfully provides a list of 
supplementary readings for those inter-
ested in exploring further the theoretical 
bases of his model. For some readers, 
such aids may not be enough to offset the 
difficult nature of the text; however, those 
who persevere will be rewarded with a 
genuinely illuminating account of IR as a 
human phenomenon.—Thomas M. Dousa, 
University of Illinois, Urbana-Champaign. 

 Kimberly Black. What Books by African 
American Women Were Acquired by 
American Academic Libraries? A Study of 
Institutional Legitimation, Exclusion and 
Implicit Censorship. Lewiston, N.Y.: Ed-
win Mellen Press, 2009. 211p. $109.95 
(ISBN 9780773437920). LC2010-278113.

The author of this volume, an Assistant 
Professor in the School of Information 
Sciences at the University of Tennessee 
in Knoxville, has produced a probing and 
thoughtful examination of the impact of 
power, and the lack of it, on two commu-
nities that she links through explication 
and analysis. The book-ended commu-
nities are (1) African American women 
writers and (2) acquisition librarians in 
academic institutions, groups who com-
municate and connect (or not) through 
filtering webs of publishers, journals, 
reviewers, and reviews. With clear prose, 
backed with statistical models, she lays 
out her arguments like a scientist, draws 
her conclusions and leaves very few open-
ings for criticism. Despite its dry title, and 
somewhat forbidding-looking tables and 
appendices (but handsome cover), the 
book offers serious rewards for readers 
interested in a variety of topics that im-
pact librarianship and our culture at large.

Dr. Black begins with a succinct sum-
mary of the distinctive traits of prose 
and mostly poetry produced by African 
American women in the late twentieth 
century, a discussion that can be of use to 
undergraduates researching that specific 
topic. She focuses her study specifically 
on the 233 titles produced by African 
American women and published by