skill and cognition in assessing mental activity from the ... · 2002 by the wenner-gren foundation...

693

C u r r e n t A n t h ro p o l o g y Volume 43, Number 5, December 2002� 2002 by The Wenner-Gren Foundation for Anthropological Research. All rights reserved 0011-3204/2002/4305-0001$3.00

Skill and Cognition inStone Tool Production

An Ethnographic Case Studyfrom Irian Jaya1

by Dietrich Stout

Stone tools represent some of the best remaining evidence of pre-historic behavior and cognition. Interpreting this evidence prop-erly requires models based on observable phenomena in the mod-ern world. For this reason, ethnographic research was undertakenamong the adze makers of the village of Langda in IndonesianIrian Jaya. This research, involving observation, interviews, andanalysis of lithic products, revealed a technology of great sophis-tication and complexity. Adze-making skill is acquired through aperiod of apprenticeship that may last five years or more, duringwhich time the community of adze makers provides a socialscaffold for the learning process. Adze production is itself a so-cial phenomenon, defined as much by personal and group rela-tions, social norms, and mythic significance as by specific reduc-tion strategies and technical terminology. Adze-making ability isassociated not only with well-developed perceptual-motor andcognitive skills but also with a wealth of technological knowl-edge. Although much of the complexity of the Langda adze in-dustry would be invisible to an archaeologist, evidence of knap-ping skill is preserved in attributes of the durable stone artifactsproduced. This evidence may be used to develop productive hy-potheses about the psychological implications of prehistoricstone tools.

d i e t r i c h s t o u t is a Ph.D. candidate in anthropology at Indi-ana University (Bloomington, Ind. 47405, U.S.A.) and a researchfellow of the university’s Center for Research into the Anthropo-logical Foundations of Technology. Born in 1972, he received aB.A. from James Madison University in 1994. His publicationsinclude “Constraint and Adaptation in Primate Brain Evolution”(Behavioral and Brain Sciences 24:295–96) and (with N. Toth, K.Schick, J. C. Stout, and G. Hutchins) “Stone Tool-Making andBrain Activation: Positron Emission Tomography (PET) Studies”(Journal of Archaeological Science 27:1215–23). The present pa-per was submitted 11 i 02 and accepted 24 v 02.

1. The research presented here was funded through generous grantsfrom the Ligabue Study and Research Center in Venice, Italy, andthe Center for Research into the Anthropological Foundations ofTechnology (CRAFT) at Indiana University. None of this workwould have been possible without the remarkable craftsmanshipand cooperation of the Langda adze makers, the assistance and com-panionship of Daniel Balyo, and the invaluable aid of Dick andMargareet Kroneman. I would like to thank Giancarlo Ligabue forhis groundbreaking work in Langda and Nicholas Toth and KathySchick for advice and assistance in the preparation of this manu-script, including contribution of some of the illustrations. Mythanks also go to Polly Wiessner for drawing my attention to thework of John Burton, Anna Stroulia for aid in translation, PaulJamison for statistical advice, Deborah Wilkerson for lithic illus-tration, and Cindy T. Cordero for proofreading. Inspiration for many

Assessing mental activity from the material residues ofbehavior is a central problem for archaeologists workingin any time period. In the case of Paleolithic archaeology,the endeavor is further weighted with evolutionary im-plications (e.g., Holloway 1969, Parker and Gibson 1979,Isaac 1986, Wynn 1993, Gowlett 1996, Ambrose 2001).Because of their persistence in the archaeological recordand because they provide evidence of individual tech-nical acts, knapped stone artifacts are commonly usedas a source of information about prehistoric and pre-modern mental abilities. Although stone tools do notnecessarily provide evidence of the full range of theirmakers’ mental abilities, they do indicate certain min-imum required competences (Gowlett 1996). In order toassess these competences, we must ultimately rely uponmodels developed from actualistic research in the mod-ern world. Since it is impossible to observe the evolvingtechnological behaviors of extinct hominid species di-rectly, our reconstructions should at least be able to drawupon an understanding of the modern human condition.

In order to contribute to the development of suchan understanding, I undertook ethnographic researchamong the adze makers (fig. 1) of the village of Langdain Indonesian Irian Jaya. Langda is located on a smallplateau in the ruggedly mountainous central highlands(fig. 2), approximately 100 kilometers southeast of thetown of Wamena in the central Baliem Valley. Althoughmetal is rapidly replacing stone for most everyday tasksin Langda, the traditional manufacture of stone adzes(fig. 3) is still practiced. These adzes, which were tradi-tionally used to clear land and to work wood, amongother functions, are produced by a semihereditary all-male community of skilled craftsmen. This communitycurrently includes both established experts and less pro-ficient or apprentice individuals. The adze makers ofLangda were first described by Giancarlo Ligabue, whovisited the plateau in 1984 with Gunter Konrad (Ligabue1985, Salvatori 1986). In 1990, Ligabue returned withDesmond Clark and Nicholas Toth to provide a moredetailed description (Toth, Clark, and Ligabue 1992). Theadze industry of Langda is also described in Pierre andAnne-Marie Petrequin’s monograph on the stone axes ofIrian Jaya (1993:217–62) and in O. W. Hampton’s (1999)Culture of Stone. All of this previous work laid the foun-dation for my own research in Langda, undertaken duringthe fall of 1999.

As was recognized long ago by the Soviet psychologistL. S. Vygotsky (1976:20–23), human tool behavior in-volves a close integration of cognitive and symbolic pro-cesses with “practical intelligence” or “technical think-ing.” The capacity for skilled technical performance is

of the questions addressed and perspectives adopted in this studycomes from the proceedings of the 1990 Wenner-Gren symposium“Tools, Language, and Intelligence: Evolutionary Implications,”held in Cascais, Portugal, and published as Gibson and Ingold(1993). Any faults with the research presented here are, however,mine alone. [Supplemental material appears in the electronic edi-tion of this issue on the journal’s web page (http://www.journals.uchicago.edu/CA/home.html).]

694 F current anthropology Volume 43, Number 5, December 2002

Fig. 1. A skilled Langda adze maker in traditionaldress.

a central aspect of human mentality that may be con-sidered alongside symbol use as a hallmark of humanity(Oakley 1954, Preston 1998). It is ultimately not the pres-ence or absence of any specific type of behavior thatembodies human uniqueness but the “generative inter-play” between tools, language, and society (K. Gibson1993).

In the words of Leroi-Gourhan (1965:35), “the tool ex-ists only with the gesture which renders it technicallyefficient.” Human technology is not a static collectionof material objects and technical facts but a dynamicsystem of skilled and goal-directed action in a social con-text (cf. Ingold 1993:433–36). One way in which to ap-preciate the nature of human skill is from an “ecologi-cal” or “perception-action” perspective. This view,arising out of the work of J. J. Gibson (1979) and N.Bernstein (1967), sees skill as a dynamic property of theorganism-in-environment system rather than as an in-ternal program expressed by the mind. Skill acquisitionis a process of learning how to act in order to solve aproblem rather than one of acquiring some rigid motorformula. The acquisition of tool-using skill in particular

involves learning to perceive the action possibilities af-forded by relations between objects (Lockman 2000)through the dynamic coupling of perception and action.

The perception-action perspective challenges tradi-tional distinctions between cognitive and motor behav-ior and points the way toward a more integrated view.In this view, thinking and doing are both behaviors ofthe organism-in-environment system; they are simplydifferent behaviors. To use the example of Thelen andSmith (1994), weaving and thinking about weaving bothemerge (in part) from patterns of neuronal activity. Thesepatterns may have partially overlapping histories andcomponents, but the critical difference in the overall en-semble remains. Neuropsychological research on mentalimagery (review in Kosslyn, Ganis, and Thompson 2001)has demonstrated just such overlap between internalconceptualization and external perception, as well ascharacteristic differences. This overlap may help to ex-plain the well-known capacity of mental practice to en-hance performance (e.g., Driskell, Copper, and Moran1994).

In any case, a human technology like weaving clearlyinvolves both weaving and thinking about weaving, aswell as talking and acting about weaving. A proper studyof technology must address this entire continuum of be-haviors. Although the traditional anthropological dis-tinction between knowledge and know-how (e.g., Hodder1990) may be overly rigid, it is important to recognizethat some kinds of goal-directed actions are more con-ceptual than others. This differentiation is particularlyimportant from an evolutionary perspective in that dif-ferences in neural substrates may also be involved.

Much of the conceptualization inherent in humantechnology relates to the fact that technology is itself aninherently social phenomenon. Stone knapping, likeother technical skills, takes place in highly structuredsocial and physical contexts (Ingold 1997) that serve toprovide “scaffolding” (Wood, Bruner, and Ross 1976) forskill learning and performance. Humans live in a con-structed environment, physically, socially, and cogni-tively. Information is not simply confined within theheads of individuals but distributed throughout this con-structed environment. A concrete example is providedby Gatewood (1985:206–7), who describes the way inwhich the design of a salmon-fishing boat on which heworked provided a “spatial mnemonic” that made it eas-ier for him to learn his job. In a similar fashion, P. Graves(1994) has pointed out that the information inherent inthe design of a bicycle is integral to the acquisition ofbike-riding skill. In the case of adze making, vital infor-mation is embodied in the spatial organization and stor-age of resources as well as in the physical characteristicsof the tools and materials used.

Skill learning and performance are also structured bysocial context, particularly with respect to such criticalparameters as task definition, motivation, and support.Instruction provided by more experienced individuals isone particularly obvious example of the role of socialcontext. Ideally, such instruction enables learningwithin what Vygotsky (1976:84–91) refers to as the zone

stout Skill and Cognition in Stone Tool Production F 695

Fig. 2. The island of New Guinea, showing the location of Langda.

of proximal development, a learning space defined bythe difference between “actual” (unassisted) and “poten-tial” (instructed/collaborative) ability. Learning occurswithin this zone as the result of the structure and mod-eling provided by the instructor, which function as a kindof scaffold for the “guided rediscovery” (Ingold 1997:111)of competence by the student.

Although somewhat more subtle than direct instruc-tion, interactions with the broader social environmentalso need to be considered. For example, the demands oflearning a difficult skill like the wheel-throwing of pot-tery may influence the technique’s distribution in dif-ferent social contexts (Roux 1990). Students of “everydaycognition” have examined some of the ways in whichsuch differences in social context are reflected in learn-ing and performance. Major factors include motivationand emotional involvement, continuity with prior ex-perience, degree of task definition, and flexibility of ac-ceptable solutions, among others (Sternsberg, Wagner,and Okagaki 1993:206). It is the influence of such factorsthat explains why, for example, some individuals per-form better on a task involving price comparison in asupermarket than on an “equivalent” paper-and-penciltask (Lave, Murtaugh, and de la Rocha 1984). In short,no task exists apart from the physical and social matrixthat gives it structure and meaning. In order to appreciatethe mental dimensions of stone knapping or any otherskill, this fact must be taken into account.

Unfortunately, this kind of detailed contextual infor-mation is difficult to derive from the fragmentary ar-

chaeological record. To exploit fully what the record cantell us, it is necessary to use interpretive models basedon observable phenomena in the modern world. One im-portant source of such actualistic data is study of (andby) modern nontraditional knappers (e.g., archaeologists,hobbyists). The nontraditional context is especially suit-able for experimental investigations of particular stone-knapping technologies (e.g., Bordes 1947, Crabtree 1966,Callahan 1979, Toth 1985), including their mechanical(Dibble and Pelcin 1995), biomechanical (Marzke et al.1998), and neurological (Stout et al. 2000) dimensions.

In an ideal world, actualistic study of nontraditionalknappers would be augmented by ethnographic researchwith diverse groups of traditional stone knappers. Suchresearch could begin to explore some of the broader con-textual variables surrounding the realization of stone-knapping skill by individuals and groups. Roux, Bril, andDietrich (1995), for example, have provided a particularlygood example of the exploration of skill in an ethno-graphic context. Unfortunately, traditional social con-texts in which to study lithic technologies are very rarein the modern world. In addition to the precocious workof Skertchly (1879) with English gun-flint makers, con-temporary reports come from a handful of groups in Ethi-opia (Gallagher 1977, Clark and Kurashina 1981, Brandt1996, Weedman 2000), India (Roux, Bril, and Dietrich1995), Australia (Tindale 1965, Gould 1980, Hayden1977), and New Guinea (Vial 1940, Strathern 1970,Burton 1985, Toth, Clark, and Ligabue 1992, Petrequinand Petrequin 1993, Hampton 1999). Shrinking this sam-


Fig. 3. A finished adze from Langda.

ple even further is the fact that the traditional manu-facture and use of stone tools in Australia appear to haveceased by at least the early 1990s (Cane 1992).

The adze makers of Langda provide a rare and inval-uable opportunity to study a skill-intensive knappingcraft in a traditional social context. Obviously, any suchcase study provides but a single example of the myriadways in which lithic technologies might be incorporatedinto modern human societies. Due caution is of courserequired in attempting to draw inferences about the pastbased on such limited evidence, and this is especiallytrue when dealing with premodern hominid species.Nevertheless, the example provided by the adze makersof Langda remains of great value. The real contributionmade by these craftsmen is not in providing answersabout the past but in helping us to frame productivequestions.

The Adze Makers of Langda

The Langda plateau is located at an elevation of 1,860meters (Hampton 1999:251) in the Ey (French: Heime)River valley, about 800 meters above the river itself. Theplateau is surrounded on all sides by soaring mountainpeaks and the precipitous slopes of deeply incised rivervalleys. Thick vegetation covers much of these slopes,and travel to nearby villages, even those within sight of

the plateau, can take the better part of a day. Even a visitto the stone quarry sites on the banks of the Ey Riverbelow Langda involves a round-trip of two and a half tothree hours. [More detailed discussion of the researcharea is presented in appendix A in the electronic editionof this issue on the journal’s web page.]

At the time that this research took place, in the fallof 1999, there were seven men living on the plateau whowere actively involved in adze making. These includedthree acknowledged experts, three apprentices, and oneolder man who was recognized as an established crafts-man but whose skill was of a lower level. The researchpresented here also includes two experts from neighbor-ing villages who visited the plateau during the studyperiod and one complete novice from Langda who hadhad no prior experience in adze making, bringing thetotal to five experts and five relatively unskilledknappers.

an overview of adze production

Stone adze heads are produced from dark gray-green boul-ders found along the banks and in the bed of the Ey River.The river flows several hundred meters east of and 800meters below the plateau at the base of a very steep slope(fig. 4). The boulders exploited by the adze makers rangein size from about 30–40 centimeters to several metersin maximum dimension.

The raw materials used by the adze makers have beenidentified petrologically as variably metamorphosed ba-salt, basalt/andesite, and metabasalt/andesite with arange of textures including fine-grained, fibrous, andophitic (lath-shaped plagioclase crystals enclosed inlater-formed augite) (Hampton 1999:84; Petrequin andPetrequin 1993:226). The quarrying of these materials isusually done in groups, with the more experienced menleading the search for materials of suitable quality. In-dividuals interviewed indicated that finding high-qualitymaterial is one of the most difficult and important as-pects of adze production.

Because of the time required for the search, quarryingactivities often continue for several days or more at astretch. During this time, the adze makers will spendnights together in huts located near the banks of the riveror along the slopes of the valley below the plateau. Boysand girls often accompany the adze makers on such quar-rying trips. During the study period, two two-day quar-rying trips were made to the Ey River, with nights spentin huts at Tukuplu and Yintong.

Quarrying begins upon arrival at the river on the firstday and when sunlight reaches the bottom of the rivervalley on subsequent days. The adze makers select a par-ticular stretch of the river on which to focus for the dayand disperse to look for promising boulders. The Ey Riveris swift-flowing, and its banks are too steep to approachit in many places; access is gained at various cobble barsor “beaches.” These locations are given names such asKaltehme Dala, Amal Dala, Ambrume Dala, and Kwa-ting Dala (dala p beach). At least some of these locations


Fig. 4. The Ey River Valley at Langda.

seem to have mythic or spiritual significance attachedto them.

When one of the craftsmen locates a promising boulder(fig. 5, top), a test flake is struck in order to reveal thebol or “deep skin” of the rock (fig. 5, center). This ex-posed surface may also be wetted down in order to makeits crystalline structure more easily visible (a techniquecalled magalingna). Craftsmen who happen to be in theimmediate area gather to discuss the merits of the se-lected boulder, commenting on the size and uniformityof the grain, the danger of internal flaws (ismar), and thepresence of black (bataya) or white (boladiatenga, “deep-skin belt”) mineral bands (fig. 5, bottom). Although theyrealize that these bands often represent points of weak-ness in the rock, the adze makers prize them for theiraesthetic value.

If a boulder is selected for quarrying, the way in whichit is attacked depends on its size (fig. 6). Often, partic-ularly if it is in the flow of the river, the boulder mustbe moved. For small boulders this presents little prob-lem, but for larger boulders a cooperative effort usingopportunistic driftwood levers is often needed. Boys whohave accompanied the adze makers may assist with thiswork.

For smaller boulders, the production of flake blanks(ya-las [ya p stone]) begins immediately. If the bouldercore (ya-dup) is small enough to be easily manipulatedby hand, blanks will be produced using a medium-sized

(∼10–15 cm, 0.5–1 kg) hammerstone (ya-winwin) and adriftwood anvil (ya-sigita). Larger boulders are attackedwith a larger (∼25–30 cm, 5–10 kg) hammerstone (da-bim). Both dabim and winwin percussors are selected inan expedient fashion from the readily available riverstones. The dabim is commonly raised over the headwith both hands and brought forcefully down on theboulder (fig. 7). Legs are spread for balance, and the flak-ing action sometimes resembles the between-the-legshike of an American football center (Toth, Clark, andLigabue 1992). Less commonly, the hammer is thrown.

For the largest boulders, fire is used to aid in fracture.After a large boulder has been sufficiently heated (some-times for most of the day), it is cracked open and thenreduced by direct percussion with a dabim. Flakes andlarge fragments produced may be used directly as blanksor, if they are particularly large, further reduced usingthe hammer-and-anvil technique described above.

As quarrying activities continue through the day, in-dividual men will alternate between blank productionand roughing out. Rough-outs (ya-temen) (fig. 8) are pro-duced from blanks through hard hammer percussion us-ing a ya-winwin hammerstone. This process is calledtemena. The core is held in the nondominant hand (inthe event, all of the men studied were right-handed andheld the cores in their left hands). A low frequency ofplatform preparation and hammerstone rubbing (dis-cussed below) is associated with roughing out.


Fig. 5. The selection (top), test-flaking (center), anddiscussion (bottom) of a potential boulder core.

Roughing out occurs both at the quarry site and at thehuts where the men spend nights between days on theriver. As with blank production, in which several menmay work the same boulder core, a single blank may bereduced by two or more men before reaching the rough-out stage. During a quarrying expedition, rough-outs ac-cumulate in piles on the beach and are stored at nightunder the eaves of the hut being used. Some of these aretaken by the individuals who produced them, usually

senior craftsmen, but most are collected and redistrib-uted by the head adze maker, who exercises authorityover the stretch of the Ey River below Langda.

When quarrying is complete, rough-outs are carefullywrapped in leaves and carried back up to the plateau innet bags. Some rough-outs are taken by individuals totheir homes, but many are stored at the hut of the headadze maker. Ya-winwin hammerstones are also storedunder the eaves of the head adze maker’s hut, and, atleast during my stay in Langda, the head adze makerinsisted that all knapping on the plateau be done in frontof his hut.

In the final phase of knapping (called talena), rough-outs are reduced to a form called a kil-ya (fig. 9) that isready for grinding and hafting. This involves an assort-ment of basic knapping operations and commonly takesabout an hour, depending on the size of the core and theskill of the knapper. It is done squatting on the ground,most often with a line of other knappers all facing in thesame direction. A premium is placed on reducing thecore as closely as possible to the ultimately desired formbecause subsequent grinding is both laborious and te-dious. As noted by Petrequin and Petrequin (1993:253),the emphasis on knapping over grinding seen in Langdaand some other areas increases productivity but also re-quires more skill than the grinding that is the predom-inant mode of reduction in other Highland axe/adzeindustries.

As knapping proceeds there is a great deal of social-izing, including discussion of the ongoing work. It is alsotraditional for adze makers to call out after a particularlysuccessful flake removal. The most common exclama-tions are “harak” and “Alim-Ey,” the latter being a ref-erence to the mythical figure Alim Yongnum, who isrevered as the provider of the tool-stone found in the EyRiver. Sometimes the flakes (ya-tokol) produced are heldaloft in display or passed along the line for examination.It is also common for knappers to observe and commenton the work of their neighbors (particularly if theseneighbors are less experienced) and even to give aid bytaking over for a while from another individual who ishaving difficulties.

Fine knapping (talena) is accomplished using an arrayof ya-winwin hammerstones kept conveniently by thefeet. These hammerstones range in mass from about 250grams to over a kilogram and represent a variety ofshapes and stone types. I was informed that hammer-stone substitution is not done in any consciously sys-tematic way (switches are made in an experimental fash-ion as difficulties are encountered), although I did notea trend toward the use of smaller percussors as core sizedecreased. This trend was also reported by Toth, Clark,and Ligabue (1992).

During knapping, the rough-out is held cradled in thenondominant (left) hand, with particularly large rough-outs being supported by the forearm as well. The fingersof the left hand are placed directly underneath the in-tended flake removal. Experts in Langda state that thisis an important part of proper technique and that thefingers serve the same function as the wooden anvils


Fig. 6. The generalized production sequence for stone adze heads in Langda.

used during quarrying (i.e., to absorb shock and directthe force of the blow). The careful placement of the fin-gers may also help in aiming. After a successful strike,the detached flake is caught and discarded with a ster-eotyped flick of the wrist. Another aspect of successfulflaking that the knappers of Langda are well aware of isthe exploitation of core morphology. Thus, skilled knap-pers look for ridges (magnana) on the core that will con-duct force and facilitate flake removal and strike at highpoints (arayna) along the edge being worked rather thanat concave flake scars (kwagna).

Prior to attempting flake removal, the knapper almostalways prepares both the striking platform (kau) and thepercussor itself. Two kinds of platform preparation arepracticed: pounding (ungna-kiringna) and microflaking(hook dongaboka). The edge of the intended platformmay be pounded with a special basalt/andesite tool(ungna), usually a chunk from a broken rough-out. Blows

are aimed directly into the body of the core and serve toremove small projections and overhangs. This is done“to keep the platform from breaking” upon percussion.Microflaking is done through a rapid succession of light,highly tangential blows with the hammerstone. The adzemakers also pay attention to preparing the surface of thepercussor they are using. This is done by rubbing (krikun)the hammerstone against the core in order to smooth thepercussor’s surface. Somewhat surprisingly, hammer-stone abrasion does not seem to be used as a techniquefor platform preparation. As was also observed by Petre-quin and Petrequin (1993:231), rubbing is done on anyflat surface of the core and is in no way focused on theanticipated striking platform. The adze makers them-selves report that the purpose of the abrasion is to preparethe surface of the hammerstone, particularly by gener-ating a small flat area that will not “slip” on contact(Hampton 1999:267). This constant abrasion (before al-


Fig. 7. An adze maker using a large hammerstone(dabim) to strike flake blanks (ya-las) from a medium-sized boulder core (ya-dup).

Fig. 8. A rough-out (ya-temen) ready for finalreduction.

most every series of blows), combined with normal wearand tear, results in heavily modified hammer-stones—with one extreme case being reduced to a near-perfect sphere.

After the finished kil-ya form has been achieved, thepiece is ground and hafted. These stages of productionare not the focus of the current investigation, but pub-lished descriptions are available in Toth, Clark, and Li-gabue (1992), Petrequin and Petrequin (1993), and Hamp-ton (1999). Briefly, the finished kil-ya is ground againsta special sandstone slab (ya-yok) obtained in trade fromthe village of Bomela, about a day’s walk to the west.Water, either from a stream or from a depression in theground, is used as a lubricant. Grinding is commonlydone in shifts and can take several hours for a large piece.The entire tool surface is ground, possibly in order toeliminate sharp projections that might damage the bind-ing. Some of the deeper flake scars are usually left intactand may be painted with red and white pigments. Thesemarkings are both decorative and symbolically mean-ingful: Petrequin and Petrequin (1993) report on infor-mants “giving life” to the adze by putting “blood” in its

wounds, while my own interviews suggest that pig-mented adzes are reserved for use by men only. Theground adze head, ready to be hafted, is known as a kil-ya-ba.

Adze handles are made from a particular variety of treelocally known as the tilyi and identified by Petrequinand Petrequin (1993:248) as belonging to the genus Gar-cinia. Adze-handle shafts (ya-kala) are made from prox-imal branch segments with the transverse haft itself con-sisting of a portion of the trunk. The blade socket is linedwith small pieces of wood (ya-suka) that act as shockabsorbers, and the blade is bound to the haft with rattan(ya-tapu).

Apprenticeship and Socioeconomic Context

Perhaps because of the high value traditionally placedon stone adzes, a good deal of social control is exertedover the materials and skills needed to produce them.As has also been reported by Petrequin and Petrequin(1993) and Hampton (1999), different villages along theEy River valley control access to quarry sites along ad-jacent stretches of the river. The head adze maker in eachvillage has personal authority over quarrying activitiesin these areas. Hampton (1999) attributes this authorityto a system of hereditary ownership.

The current head adze maker of Langda regulates the


Fig. 9. A finished adze-head (kil-ya) produced by a skilled craftsman, along with representative shaping flakes.Note the long, narrow flake scars emanating longitudinally inward from the bit of the adze head, a characteris-tic feature left from the removal of specialized blade-flakes (utina).


quarrying process by collecting and redistributing therough-outs produced during trips to the river. This pro-cess of redistribution is not a rigorous one, and men areoften allowed to take home the same pieces that theyproduced, but the recognition of authority is clear. Backon the plateau, hammerstones and rough-outs are storedat the home of the head adze maker. He enjoys rights ofownership over these materials, and I was informed thatin the past men had been killed for using such itemswithout permission.

Some of the influence wielded by the head adze makerof Langda is almost certainly due to his own personalityand charisma. As noted by Petrequin and Petrequin(1993) and Hampton (1999), he is both the head adzemaker and one of the most influential headmen in hiscommunity, a combination that is by no means univer-sal. Be that as it may, his authority also rests on tradi-tional and hereditary arrangements that are consistentwith the broader social and environmental context. Thismay be illustrated through comparison with the tradi-tional adze industry (described by Petrequin and Petre-quin 1993) of the nearby Phu River valley, locatedroughly 40 kilometers west-north-west of Langda.

In the past, inhabitants of the Phu Valley followed acycle of itinerant agriculture that brought them near tosources of adze stone for roughly one month out of theyear. These sources, large detrital cones along the PhuRiver, were treated as communal resources open for ex-ploitation by all. Rights of ownership to stone blocksand large flake blanks simply went to the individual whocollected them. Expert knappers were paid, usually withvegetables or pork, to work these blocks into a formready for grinding, but ownership of the stone remainedwith the original collector. This is obviously very dif-ferent from the situation in the Ey River valley, wherehereditary quarrying rights are closely guarded by com-munities and individuals.

There are many potential explanations for this differ-ence, including stochastic historical variation, but onemajor candidate is the difference in mobility and sea-sonality between the two groups. In Langda, raw-mate-rial sources are close at hand throughout the year, andadze manufacture may take place at any time. This fa-vors the development of safeguards against overproduc-tion and overexploitation and allows individuals to mon-itor access to quarry sites. In the Phu Valley, in contrast,lithic resources are only at hand for a small part of theyear. This produces a situation in which not only mustproduction be maximized for a short period but contin-uous monitoring of quarry sites is essentially impossible.In each case, the broader social and environmental con-text is reflected in the organization of the local adze-making industry. Although some of the authority exer-cised by the head adze maker in Langda may result fromthe force of his personality, a good part also derives fromthe traditional arrangements that have developed in theparticular environmental context of Langda.

Another aspect of the adze industry in Langda that issubject to social control is skill itself. Entry into thecommunity of adze makers occurs through a period of

apprenticeship that can last five years or more. TheLangda craftsmen report that, because of the great valueof the skill, they will instruct only close relatives, usu-ally “sons” (the Langda kinship system does not differ-entiate between sons and nephews). In general, a poten-tial apprentice will express interest by attending to theknapping activities of the craftsman, perhaps bringingfood or helping with menial tasks. For his part, the po-tential master will attempt to evaluate the seriousnessand commitment of the student. He may even want tojudge some early attempts at knapping before decidingto accept the apprentice.

Traditionally, apprenticeship began around the age of12–13. In Langda today, however, the apprentices are gen-erally in their mid-twenties. This is probably a reflectionof the changes in the adze-making craft that have re-sulted from the increasing availability of metal tools.Demand for adzes to be used in everyday labor has fallenoff dramatically, reducing the economic value of adze-making skill. For the adze makers, however, the craftremains as a source of pride and social identity. It seemslikely that the changing significance of the stone adzein the social and economic life of Langda has altered themotivations for skill learning and caused more individ-uals to make the choice to pursue apprenticeship laterin life.

Despite this shift, many traditional aspects of appren-ticeship remain. Some are restrictive, such as the factthat any profits (in money, food, or goods) made by theapprentice must be given to the master. The apprenticeis similarly restricted in the size of the rough-outs he isallowed to work on. During my research, I repeatedlyattempted without success to persuade the head adzemaker to allow an apprentice to reduce a large rough-out for the purpose of comparison. Such restrictionsprobably help to limit the potentially disruptive influ-ences of novices (B. Graves 1989) by preventing com-petition with established craftsmen and regulating theuse of valued resources.

Restrictions also play an important role in the learningprocess. For example, I was informed that it can take tenyears or more to develop the skill needed to produce oneof the largest adzes, which reach over 270 millimetersin length. In contrast, apprentices often produce piecesthat are too small (∼100 millimeters) ever to be used andare made solely for practice. Forcing apprentices to prac-tice on small rough-outs suited to their level of experi-ence helps to situate learning within an appropriate zoneof development.

Social scaffolding for the learning process is also pro-vided by the dynamics of the adze-making communityitself. Virtually all technical operations in the productionof stone adzes are conducted as group activities with agreat deal of interaction among individuals. This in-cludes discussion, observation, demonstration, and evendirect assistance. It is quite common, for example, forone man to make suggestions to another about where toattempt the next flake removal or to comment on thequality of the material being worked. Similarly, a workerwho is experiencing difficulties may ask another to try


Fig. 10. Two experienced knappers (first and secondfrom right) providing instruction to an apprentice(third from right).

Fig. 11. Some of the terms used to identify differentparts of an adze and different kinds of flakes. Transla-tions are in parentheses; note that many parts are de-scribed using anatomical metaphors.

his hand at the task. Such interaction occurs betweenpeers but is most frequent between experts and appren-tices (fig. 10). Experts also engage in more general ex-position about proper stance, technique, and even atti-tude. Observed examples include apprentices’ being toldto work more slowly, to keep in mind the way in whichthe adze will be hafted while shaping the butt, to thinkbefore striking, and to strike the “high” or projectingparts of an edge first. Apprentices benefit from demon-stration, instruction, assistance, and motivation fromthe entire community of craftsmen, not simply fromtheir own nominal masters. Because of this communaldynamic, apprentices are able to participate in all stagesof production, even those that they would be unable toaccomplish by themselves. Furthermore, they are givenassistance and advice during each stage that allows themto perform well beyond the unassisted level. This socialfacilitation of precocious performance is a practical re-alization of Vygotsky’s zone of proximal development,in which structured support allows students to acquireskills more rapidly.

Social context also provides motivation and produceswork-related values. A central value expressed by theadze makers is commitment to and immersion in thecraft. Thus, I was repeatedly told that proficiency is ac-quired only through continual practice and that withoutit a man’s hand will “grow heavy.” Similarly, craftsmencompliment others by commenting on their single-minded commitment to knapping and speak of forgettingtheir wives, their gardens, and everything else while theyare working. In addition to the traditional exclamationsof “harak!” and “Alim-Ey!” while working, the crafts-men express their joy in their work by speaking of theflakes as “peeling off like sweet-potato skin” or slidingoff as if on the “wet grass of a mountain slope.” Oneman spoke excitedly of wanting to flake “every stone inthe river.” In sum, apprenticeship in Langda is a socialinstitution that reflects the broader socioeconomic con-

text of the adze-making craft while at the same timecontributing significantly to the structured social envi-ronment in which learning takes place.

Conceptualization and Communication

Adze making in Langda is associated with a large bodyof terminology and other technological concepts. For ex-ample, stone adze heads are subdivided into many dis-tinct parts identified by particular names (fig. 11). Sim-ilarly well-developed terminology exists to describeparticular knapping techniques and strategies. Thismakes it relatively easy for the adze makers to com-municate about the details of their craft, for example,when giving technical instructions to apprentices. Thus,an apprentice may be reminded to work edges bifacially(dinar dina) and to strike at high points (arayna) ratherthan concave flake scars (kwagna) so that the flakes re-moved will tend to follow ridges (magnana) on the core.

Such statements provide a window on the consciousreduction strategies pursued by craftsmen. For example,several experts told me that for large adzes it is partic-ularly important to start work on the bit (si) first, thenthe butt (bumyok), and finally the middle section (tingkwiribkun). This is because premature thinning of themiddle makes the piece more likely to break as the bitand the butt are shaped. It was also emphasized thatspecial attention needs to be paid to shaping the dorsalridge (amyok), because it is one of the most difficult areasto flake and often gives apprentices trouble (in the re-duction strategies employed by skilled knappers this


ridge is itself a third flaked edge and not simply an in-tersection of flake scars). Such comments reflect on theconceptual “thinking about knapping” that is as mucha part of the Langda stone adze technology as actualknapping.

Further information about conceptualization comesfrom narratives describing particular reduction se-quences. On one occasion, for example, an older crafts-man explained to me that he had struck a flake from theside (abum) of the bit in an attempt to remove a stepfracture (kotangna) left during a previous flake removal.This attempt had failed, he said, but he was then ableto remove the step by striking a long blade-flake (utina)from the front of the bit. Such examples illustrate notonly the adze makers’ detailed knowledge of the mate-rials, techniques, and strategies of their craft but alsothe problem solving and higher-level organization itinvolves.

knowledge of raw materials

A more complicated example of conceptualization andcommunication is the naming of adzes. Although theLangda adzes are all made in volcanic basalts/andesitesof varying texture, the craftsmen use at least 14 differentnames to describe their lithic raw materials. In additionto these stone-type names, each adze head also receivesan ancestor name and a place-of-origin name.

Stone-type names are generally assigned after carefulexamination of the stone’s physical properties, usuallyin the context of group discussion. Naming first occursat the quarry site, but the name may change as workproceeds. I was informed that grinding in particular re-veals qualities helpful in naming an adze, such as thesize, color, and uniformity of crystals or the presence ofmineral bands. Such bands are considered to be impor-tant “signs” of a stone’s name. The naming of hammer-stones also seems to reflect stone characteristics such ascolor rather than size or shape, but additional researchis needed on this topic.

The physical characteristics consulted in assigning aname to adze-stone are almost certainly associated withtechnically relevant variables such as ease of flaking andultimate durability. In fact, I was informed that the two“special” or bikrobya types, murkonye denlaru andlumkerye, are valued specifically because they arestronger. It was also stated that these quality materialsare much more difficult to find and cannot always beidentified by the unskilled. In the observed quarryingactivities, it was in fact the more experienced men whowere responsible for locating most of the materials. Evenfor these experienced men, however, the naming of par-ticular stones is not a trivial task.

When I asked three experienced men to name a ran-dom assortment of flaked pieces independently, theyagreed in only 2 cases out of 11. This is not surprisingconsidering that naming is normally part of the processof finding and working the stone and involves substantialinterplay between individuals. The almost total lack ofagreement observed in this artificial context suggests

that naming, like so many other aspects of technologicalperformance, is an inherently social phenomenon.Burton (1985) reports a similarly elaborate taxonomy ofraw materials among the stone axe makers of the WahgiValley in Papua New Guinea. As in Langda, “no twoinformants gave the axe varieties in exactly the sameorder” (p. 60), and, when given museum specimens toidentify, “they were too inconsistent in their identifi-cations for the basis of the classifications to be clear” (p.61). These parallels strongly suggest that the naming ofraw materials was widespread in traditional Highlandsadze/axe technologies and that it was often a social andsymbolic rather than a purely classificatory process.

knowledge structure and context

As is illustrated by the naming of adzes, technologicalknowledge among the Langda adze makers is structuredin a fairly complex way. One might be tempted to dealwith this by distinguishing between “practical” tech-nological knowledge of things like the physical proper-ties of stone and more “symbolic” or socially relevantaspects of meaning. Such a distinction would, however,be somewhat artificial, being based on typically Westernideas regarding the separation of society and technology(Ingold 1997) rather than the understandings of the adzemakers themselves. For these craftsmen the productionof adzes is itself a social phenomenon.

This may be seen on several levels. For one thing, theadze makers view the stones they work with as living,intentional subjects. Thus, knappers will speak of stones’being “angry” if they fail to fracture as desired and willcall out to them using their “secret names” as theysearch for them at the quarry sites along the river. Theboulders themselves are believed to grow with age aspeople do (Petrequin and Petrequin 1993:226), with “oldstone” (wisy-ya) being darker and stronger than “youngstone” (ya-babau). Social relations with stone are an im-portant part of production, and care must be taken toavoid angering pieces through improper practices suchas placing finished pieces on the ground in an improperorientation (they should lie parallel, with the bit facingaway from the craftsman). Similarly, Toth, Clark, andLigabue (1992:92) report that when an adze breaks “themakers say they ‘feel sorry’ for their handiwork and takepains to bring it home for final discard.”

Social relations with people (living, dead, and mythi-cal) are also an important part of production. The firstthing, entirely unsolicited, that adze makers generallywanted to relate during interviews was a list of theirancestors who had handed down the craft through theyears. For the dominant Balyo clan, this list begins withMenminy Malyoman Balyo, who is said to have origi-nated the technology. Another important figure, as wehave seen, is the mythical woman Alim Yongnum, who“gives birth” to the stones in the Ey River and controlstheir availability (Louwerse 1990).

Relations with living individuals are embodied in thelearning, instruction, cooperation, and exchange that areas much a part of the adze-making craft as is knowledge


of reduction strategies. Craftsmen must always be awareof their relationships with others and the ways in whichthey are enacted and modified through adze production.The social, symbolic, and mythic ramifications of theindustry are for them in no way external to the centralgoal of adze making. Knowledge in these spheres is oneaspect of an overarching structure of knowledge that isunified by its practical and teleological focus.

Research on the subject of “everyday” or “situated”cognition has revealed that motivation, emotion, andtask definition are important components of humanmental performance. Keller and Keller (1996:23) elabo-rate this practice perspective with their concept of pro-duction constellations organized under goal-definingumbrella plans. Each such constellation embodies a“unit of ideas, tools, and materials” oriented toward theachievement of a particular step in production. In thecase at hand, the umbrella plan is the production ofadzes, and it is composed of individual constellationsstructured around raw-material procurement, roughing-out, grinding, and so on. In each of these constellations,information is distributed throughout a system that in-cludes the affordances (J. J. Gibson 1979) of the physicaland social environment as well as the performance char-acteristics (Schiffer 1999) of the craftsmen. In the raw-material procurement constellation, for instance, under-standings about desirable stone characteristics, mythicsignificance, resource ownership, and extraction tech-niques are realized through purposeful interaction withthe actual locations, stone resources, and individuals andgroups involved in quarrying.

Manifestations of Skill

To evaluate variation in knapping skill, it is necessaryto have a standard by which to judge performance. In thecurrent study, general information about norms andstandards provided by informants was augmentedthrough analysis of actual knapping products. It was hy-pothesized that performance would vary with experienceand that the adze heads produced by established crafts-men would reflect a higher level of skill than those pro-duced by apprentices. In order to test this hypothesis,individual craftsmen were divided into two groups(“skilled” and “unskilled”) according to social status (es-tablished versus apprentice). This simple division doesnot capture individual variation within groups, espe-cially in terms of the developing skills of the apprenticecraftsmen, but it does reflect large-scale patterns ofdifference.

Statistical comparison of the two groups revealed aclear pattern of differences between established and ap-prentice craftsmen, including significant differences inthe size and shape of the adze heads and waste flakesthey produced as well as in the organization of theiroperational sequences. One of the most robust and di-agnostic differences was the greater mean size (lengthand weight) of adze heads produced by established crafts-men. This observation fits nicely with interview data

indicating that larger adze heads are considered to bemore difficult to produce and are more highly valued.

There is, however, one major exception to this pattern.In the field, it was noted that one established craftsmanappeared to be performing at a lower level than his peers(i.e., making fewer, smaller, and less well-formed adzeheads). Examination of the quantitative data confirmedthis impression, revealing significant differences be-tween the adze heads and flakes produced by this indi-vidual and those from the rest of the “skilled” sample.Although the pattern of differences described above isrobust enough to be discerned whether or not this in-dividual is included in the “skilled” sample, he is clearlyan outlier from that group. In fact, he was found to dis-play the same pattern of performance as the apprenticeknappers and did not differ significantly from the “un-skilled” sample in either adze head or flake metrics. Theexplanation for this interesting exception is unknownand may include loss of skill with advanced age, a lackof innate ability, and/or some unknown social dynamic.In light of this uncertainty and because actual perform-ance was considered more directly relevant than socialstatus to the questions being addressed, the individualin question was ultimately classified as “unskilled” inthe analyses presented below.

As we have seen, adze production involves multiplestages and associated constellations. The informationpresented here will focus on final reduction from therough-out to the finished form ready for grinding.

kil-ya: products of skill

In the Langda adze industry, the ultimate objective ofstone knapping is to produce well-formed adze heads thatare ready to be ground (to become kil-ya-ba) and thenhafted with a minimum of effort. Not surprisingly, it isin the achievement of this central goal that some of themost striking differences between skilled and unskilledcraftsmen emerge.

As part of the observation of the knapping process, coremeasurements were taken both before and after final re-duction. This yielded a sample of 25 finished adze heads(prior to grinding), including 18 made by five skilledcraftsmen and 7 made by five unskilled craftsmen. Be-cause of the relatively small size of the sample, mea-surements from different adze heads made by the sameindividual were treated as independent data points. Useof only a single adze head per individual produces qual-itatively similar results but neglects many of the avail-able data. [See appendix B in the electronic edition ofthis issue on the journal’s web page.] Use of mean valuesfor individuals would eliminate the variance that is theobject of study in the first place.

Seven linear dimensions, four angular dimensions, andweight were recorded for each core. Linear measurestaken included the length along the midline of the longaxis and both width and thickness at each of three pointsalong this length: (1) at the shoulders of the bit, (2) inthe middle at half the length, and (3) at the butt at nine-tenths the length. Angular measures were taken of the


table 1Absolute Metrics for kil-ya Produced by Skilled and Unskilled Knappers

Measure and Sample N Mean Range Standard DeviationSignificance of Differ-ence between Means

Weight (g)Skilled 17 300 125–620 130Unskilled 7 203 146–240 35 0.010∗∗

Length (mm)Skilled 18 201 136–271 40Unskilled 7 150 123–165 14 0.0005∗∗

Width at head (mm)Skilled 18 43 34–52 5Unskilled 7 41 36–48 5 0.299

Width at half length (mm)Skilled 18 35 32–41 3Unskilled 7 38 31–41 4 0.124

Width at butt (mm)Skilled 18 21 19–28 3Unskilled 7 23 21–26 2 0.100∗

Thickness at head (mm)Skilled 18 20 14–39 5Unskilled 7 19 17–22 2 0.589

Thickness at half length(mm)Skilled 18 28 20–36 5Unskilled 7 25 20–30 4 0.098∗

Thickness at butt (mm)Skilled 18 15 8–22 3Unskilled 7 15 12–21 3 0.848

∗ Significant at 90% level.∗∗ Significant at 95% level.

bit angle of the blade, the slope of left and right sides athalf the length, and the point of the butt. Comparisonsof absolute and relative (ratio) values between thesegroups were made using Pearson’s two-tailed t-tests,while the relationships of variables within the groupswere explored using simple linear regression.

The most striking differences between adze heads pro-duced by the skilled and unskilled groups are the greatermean length and weight of the skilled pieces (table 1).Even from casual observation it is obvious that estab-lished experts routinely produce larger adze heads thanapprentices (fig. 12). Whereas the longest adze head pro-duced by the unskilled group measured 165 millimeters,roughly 80% of the adze heads produced by skilled crafts-men were greater than 170 millimeters in length.

Of course, apprentices are allowed to work only withrelatively small rough-outs, usually of less than 200 mil-limeters in length. In contrast, skilled workers may startwith rough-outs as long as 300 millimeters or more.Expert adze makers assert that the production of largeradzes is simply beyond the ability of apprentices, but itwas impossible to test this directly because no instancesof apprentices’ attempting to work larger cores were per-mitted. Some indirect support for the assertion is pro-vided by differences in shape between adze heads madeby the skilled and the unskilled groups. Although theformer are significantly longer, they are not thicker orwider. In fact, when width and thickness measures areexpressed as proportions of length, their mean values are

significantly less for skilled pieces at all measured points(fig. 13), despite the fact that there are no significantdifferences in the proportions of the rough-outs used byapprentices and experts.

The shape differences between adzes made by theskilled and the unskilled groups are not simply an al-lometric artifact of size. When width at midpoint is plot-ted against length (fig. 14), it is clear that, whereas skilledcraftsmen are able to maintain a constant relationshipbetween length and width across a large range of sizes,these dimensions are completely unrelated in pieces pro-duced by the unskilled group.

These observations indicate that less developed knap-ping skills are reflected in an inability to control theproportions of the finished product even if it is small andthat this problem arises mainly from difficulty in nar-rowing the piece. The inability of unskilled knappers toshape even small adze heads strongly supports the hy-pothesis that they would be unable to work larger pieceseffectively.

Explanations for this difference in ability may besought at different levels. For example, there is the globalreduction strategy employed by less skilled workers, whodirect little or no attention to flaking along the dorsalridge of the piece. Differences in motor skill may alsohelp to account for variable success in individual flakeremovals and in the ultimate achievement of knappinggoals. Exploring the relative contribution of such inter-


Fig. 12. Finished adze heads produced by skilled (three on left) and unskilled (three on right) craftsmen. Thepieces produced by the skilled group are not only longer and relatively narrower but also more regular in planform.

related factors is a major step in determining the natureof adze-making skill in Langda.

strategic skills

At the highest level of organization, there are the overallstrategies that guide reduction. Some aspects of strategyare embodied in general rules of thumb such as the needto shape the bit first or to work evenly along the edgesof the piece. These general rules do not, however, en-compass the complexity and variability of the strategicdecisions that must be made when working idiosyncraticstone cores (Pelegrin 1993). In order to gain a more de-tailed appreciation of the strategic choices made by theknappers of Langda, we must carefully examine the ac-tual patterns of technical performance through whichthese choices are enacted.

Video recordings were made of 18 individual sequencesof reduction from rough-out to finished form. Of these,12 were selected as being suitable for analysis (relativelyfree of interruptions, piece successfully worked to com-pletion, work done by a single individual, etc.), resultingin a sample of 5 reduction sequences by four skilledknappers and 7 sequences by five unskilled knappers.The question of independence once again arises, but re-

duction of the sample to only one sequence per knapperdoes not meaningfully alter the results.

Each recording was carefully reviewed and scored forthe occurrence, sequence, and location on the core of thefive major technical operations employed by the Langdaknappers: (1) talena, a forceful blow with a ya-winwinhammerstone, directed into the body of the core andintended to detach a flake; (2) ungna-kiringna, poundingof an intended platform with a small basalt block (ungna)to remove irregularities and projections; (3) hook don-gaboka, light, tangential strokes with a ya-winwin ham-merstone, used by experts to prepare platforms throughmicroflaking; (4) krikun, rubbing of the ya-winwin ham-merstone against the core to smooth the hammerstone’ssurface; and (5) dung d’ona, refitting of detached flakesand fragments onto the core (usually done for inspectiononly but occasionally used to complete the removal ofsnapped flakes using the proximal end as a punch). Inorder to score the location of an action, cores were di-vided into the five regions recognized by the knappersthemselves (fig. 5), namely the bit (si), sides (ting kwi-ribkun, including left and right together), butt (bumyok,posterior fifth of the piece), and dorsal ridge (amnyok).Elapsed time was recorded at five-minute intervals withsome deviations reflecting continuity or pace of work.


Fig. 13. Relative proportions of adze heads (scaled tolength) produced by skilled (top) and unskilled (bot-tom) craftsmen. The significance (p) of the differencein proportions for each measured dimension is indi-cated alongside the lower diagram.

The data produced were once again divided into “skilled”and “unskilled” groups; trends and differences were ex-plored using simple bivariate (two-tailed Pearson’s) cor-relation and Student’s t-tests. The significance of differ-ences in the variability of data between the two groupswas assessed using Levene’s test for equality of var-iances.

Rules of thumb regarding the sequence in which dif-ferent parts of the adze should be shaped are only weaklyexpressed in actual patterns of reduction activity. Amongskilled knappers, the general rule that the bit should beshaped first is reflected by a significant but weak nega-tive correlation between knapping time elapsed and the

proportion of blows directed at the bit (r p �0.401,p p0.031). Similarly, the mean percentage of blows directedby experts at the sides is significantly greater (p p 0.049)during the last quarter of the reduction sequence (53%)than during the first three (37%). The only such differ-ence among unskilled knappers is an increase in the pro-portion of blows directed toward the butt during the finalquarter of the reduction sequence (p p 0.057). In thiscase, the actual knapping practices of unskilled individ-uals are at odds with the expressed rule that the bladeand butt should be finished before the middle section.

There is thus some evidence for differences in the tem-poral organization of knapping strategies in skilled asopposed to unskilled craftsmen. However, this evidenceis relatively weak, and many potential comparisons notreported here do not show such patterning. It appearsthat, for the most part, decisions regarding which partof the core to work on at any particular time are char-acterized more by homogeneous variability than by ad-herence to any uniform sequence.

Strategic differences between skilled and unskilledcraftsmen are more clearly visible in the global pattern-ing of blows directed against different parts of the adzeover the entire course of reduction. Most obvious is thealmost complete lack of attention paid by apprentices tothe dorsal ridge (fig. 15). The difference between expertsand novices in the percentage of dorsal-ridge strikes ishighly significant (p ! 0.005), with skilled workers av-eraging 24% and unskilled workers a mere 2%. Con-versely, unskilled workers directed an average of 66% oftheir blows to the sides of the cores while skilled workersaveraged only 44% to this region (p p 0.060).

As may be seen in figure 15, the strategies pursued byunskilled knappers are far more variable than the rela-tively stable strategy of skilled knappers. This is furtherreflected in the higher standard deviations for unskilledworkers in every region except the uniformly neglecteddorsal ridge. The lesser variance of skilled data in eachof these regions is significant at p p 0.100 or better.Although skilled workers do not work on the differentparts of the core in an immutable sequence, they domake uniform decisions about how much overall atten-tion to devote to each region. This is not the case forunskilled workers, whose global strategies show little ifany consistent pattern.

These strategic differences are best understood as in-dicative of different levels of skill. Skill emerges fromthe dynamic coordination of perception and action withrespect to organismal, environmental, and task-relatedconstraints (Newell 1996). Skill learning can occur ondifferent spatiotemporal scales and may thus be char-acterized as more or less “conceptual” in nature, despitethe fact that it always concerns goal-oriented action.Skilled knappers in Langda learn to coordinate large-scale patterns of perception and action in their treatmentof the various parts of the core. Overall strategic regu-larity emerges from knapping patterns that are neces-sarily variable on a smaller scale and reflects the per-ception of larger relationships in the emerging form of


Fig. 14. Length versus width for adzes produced by skilled (solid line) and unskilled (broken line) craftsmen.Numbers indicate the individual craftsmen responsible for each adze head. R2(skilled) p 0.7098, (unskilled) p0.0002.

the adze. Unskilled knappers are still exploring theserelationships and fail to exhibit such strategic regularity.

Experts display an understanding that, although em-bodied in practical action, may be considered conceptualin scope. Depending on the spatiotemporal level of de-scription adopted (Keijzer 1999), an active knapper maybe said to be either moving his arm, removing flakes,shaping the dorsal ridge, making an adze, or attemptingto increase his wealth and prestige. In fact, he is doingall of these. Actions such as the pursuit of an overallknapping strategy take place toward the higher end ofthis scale and are more closely analogous to what areconventionally termed “cognitive” behaviors or skills.There is even evidence that such large- versus small-scale actions tend to involve different neuroanatomicalsubstrates (Green 2001).

Obviously, these different levels of organization areintimately (though not simply nor linearly [Keijzer 1999])related to each other. For example, the basic physicaldifficulty of striking flakes from the dorsal ridge of thecore, which requires the use of large and even obtuseexterior platform angles, almost certainly plays a role inapprentices’ neglect of this region. It may also be thatthe propensity of unskilled individuals to strike less in-

vasive flakes (see below) makes it difficult for them toachieve a workable angle along the dorsal ridge.

perceptual-motor skills

Because of the essential continuity between the cogni-tive and the perceptual-motor skills involved in adze pro-duction, it is neither possible nor desirable to dichoto-mize the two. It is, however, beneficial to differentiateskills along a continuum from small-scale perceptual-motor actions to large-scale cognitive/conceptual ac-tions. To investigate the perceptual-motor end of thiscontinuum, I collected and analyzed flake waste fromnovice and experienced craftsmen.

A 100% collection was made of the debitage producedduring observed reduction sequences. Waste productsfrom different individuals and from different coresworked by the same individual were collected, analyzed,and recorded separately. All pieces were counted, andthose greater than 20 millimeters in maximum dimen-sion were classified and analyzed. The sample presentedhere consists of all of the whole flakes from the finalreduction (rough-out to finished adze head) of 14 cores.This includes 453 flakes from 8 cores made by the skilled


Fig. 15. Percentage of total blows directed against various parts of the core during six individual reductionsequences by three unskilled and three skilled knappers.

group and 189 flakes from 6 cores made by the unskilledgroup.

Numerous attributes were recorded for each wholeflake. Ratios between overall flake dimensions (length,breadth, thickness) and between platform dimensions(breadth, depth) were also calculated in order to producedimensionless shape variables. A principal-componentsanalysis using Varimax rotation was conducted in orderto identify the major dimensions of variability withinthe overall sample; component scores for the skilled andunskilled samples were then compared using Student’st-tests.

Principal-components analysis using 12 flake size andshape variables yielded five components accounting for84% of the observed variance (table 2). On the basis ofthe factor loadings obtained, these components may bedescribed as corresponding to (I) absolute size, (II) relativethickness, (III) relative elongation, (IV) platform shape(relative depth), and (V) platform angles. T-tests revealedsignificant differences in component scores betweenflakes made by skilled and unskilled groups on compo-nents I, II, III, and V. Inspection of the mean score valuesfor each component in light of their factor loadings re-veals the direction of these differences. For example,component V (platform angles) loads positively on in-ternal angle and negatively on external angle, so that thelesser mean value of skilled flakes on this component

indicates smaller internal angles and larger external an-gles (i.e., both are closer to 90�). Overall results indicatethat the flakes made by the skilled group are absolutelylarger, relatively thinner, and more elongated and havesteeper platform angles than those made by the unskilledgroup. These interpretations may be corroborated in amore direct fashion through a careful inspection of pair-wise comparisons between the two samples (table 3).

Multiple factors must be considered in order to un-derstand the observed differences in flakes produced byskilled and unskilled craftsmen. To begin with, inten-tional production of larger, thinner, and longer flakeswould be useful for the exercise and maintenance of con-trol over core form. Longer flakes allow access to coreareas farther from a working edge, while relatively thin-ner flakes allow even shaping of an edge withoutgouging.

Before accepting this intentional explanation, how-ever, we need to consider the potential constraining ef-fects of core size. Experts tend to work larger cores, andthis might plausibly contribute to differences in flakesize and shape. In fact, component I (flake size) is cor-related with core length (p p 0.004), but this accountsfor very little of the observed variation (r2 p 0.06). Thesame is true of components II (thickness) and V (platformangles). Components III (elongation) and IV (platformshape) are not correlated with core length. Core size ex-


table 2Principal-Components of Variation in Flake Attributes

Components

VariablesI

Absolute SizeII

Relative ThicknessIII

ElongationIV

Platform ShapeV

Platform Angles

Flake dimensionsLength 0.766 �0.022 0.584 �0.038 �0.015Breadth 0.909 0.013 �0.280 �0.038 �0.038Thickness 0.706 �0.621 �0.020 �0.148 �0.018Maximum 0.948 �0.020 0.114 �0.064 �0.029

Platform dimensionsBreadth 0.663 �0.100 �0.350 0.114 0.104Depth 0.526 �0.205 �0.177 �0.625 0.132Exterior angle 0.030 �0.143 0.072 0.305 �0.804Interior angle 0.038 �0.159 0.077 0.217 0.803

Flake shapeLength/breadth �0.146 �0.051 0.932 �0.017 0.025Length/thickness �0.134 0.788 0.534 0.110 �0.060Breadth/thickness 0.020 0.927 �0.256 0.132 �0.036

Platform shapeBreadth/depth 0.069 0.125 �0.092 0.888 0.020

erts a weak influence on some flake attributes but doesnot account for the overall pattern of differences betweenthe flakes made by skilled and unskilled knappers.

Skilled knappers produce larger, flatter, and more elon-gated flakes with more oblique platform angles but with-out any significant difference in platform depth or thick-ness. According to the controlled fracture experimentsof Dibble and Pelcin (1995), flake mass increases deter-ministically with platform thickness. Looking to the cur-rent data, we find that platform thickness is closely cor-related with platform depth (r2 p 0.935, p ! 0.005) andthat both are correlated with flake thickness (r2 p 0.350,p ! 0.005 and r2 p 0.332, p ! 0.005, respectively). Themechanical properties of the stone medium thus allowskilled knappers in Langda to produce longer and broaderflakes without increasing thickness by choosing steeperexterior platform angles while holding platform depthconstant.

This strategy imposes an added burden on the percep-tual-motor skills needed for flaking. As observed by Dib-ble and Pelcin (1995:437–38), increased exterior platformangles allow the production of larger flakes but requiregreater control on the part of the knapper. More specif-ically, experts are faced with the need to increase boththe magnitude and the accuracy of the striking momen-tum they employ. According to Fitts’s (1954) law of psy-chomotor behavior, such an increase in striking mo-mentum without a corresponding increase in target sizeproduces a logarithmic increase in task difficulty. Theflakes produced by skilled knappers in Langda thus pro-vide evidence of increased task difficulty and of corre-spondingly greater perceptual-motor skill.

All other things being equal, the experts’ greater per-ceptual-motor skill would allow for the production ofrelatively thinner flakes. However, all other things arenot equal. Core morphology plays a large role in deter-mining the shape of any particular flake that is detached.

The experienced knappers of Langda are well aware ofthis and will make specific comments about phenomenasuch as the tendency of flakes to propagate along ridgeson the core. It is quite likely (though difficult to test)that one of the major factors underlying the differencesbetween the flakes (and finished adze heads) of skilledand unskilled knappers is the greater ability of expertsto exploit and manipulate core morphology.

One important example is the blade-flakes (utina) (fig.16) produced by experts while shaping the bit of the adzehead. Even though these flakes are not significantly dif-ferent from others in platform shape (breadth/depth) orexterior angle, they are roughly twice as elongated (flakelength/ breadth p 2.0 compared with 0.9). This differ-ence is achieved by manipulating core morphology inorder to create long ridges running perpendicular to thebit, a technique that is broadly analogous to the prepa-ration of a blade core (e.g., Whittaker 1994:119–21). Theridges produced serve to guide the propagation of flakesinward, preventing the lateral dissipation of force.

The diagnostic blade-flakes that are produced in thisfashion are almost entirely absent from the debitage pro-duced by the unskilled group. Although these blade-flakes are an extreme example, they do indicate that theselection of advantageous targets and the manipulationof core morphology are important components of theknapping skill demonstrated by experts. In one sense,such technical choices are made on a relatively concep-tual level of organization, but they are also grounded inthe perceptual-motor skills needed to realize concreteplans in a difficult stone medium.

Discussion

Adze making in Langda is a skilled technical activitythat exists as one aspect of a broader socioeconomic


table 3Pair-wise Comparison of Attributes of Flakes Produced by Skilled and Unskilled Knappers

Attribute and Sample N Mean Student’s t-test p≤ Kolmogorov–Smirnov p≤ Mann–Whitney p≤

Length (mm)Skilled 153 30.2Unskilled 189 25.7 0.0005∗∗ 0.0005∗∗ 0.0005∗∗

Breadth (mm)Skilled 453 33.7Unskilled 189 31.3 0.042∗∗ 0.146 0.025∗∗

Thickness (mm)Skilled 453 4.9Unskilled 189 4.7 0.446 0.647 0.821

Maximum dimension (mm)Skilled 453 42.7Unskilled 189 37.8 0.0005∗∗ 0.0005∗∗ 0.0005∗∗

Platform breadth (mm)Skilled 438 17.8Unskilled 186 15.5 0.004∗∗ 0.021∗∗ 0.002∗∗

Platform depth (mm)Skilled 444 3.4Unskilled 188 3.4 0.964 0.614 0.592

Platform thickness (mm)Skilled 421 3.4Unskilled 186 3.2 0.292 0.573 0.661

Exterior platform angle (�)Skilled 422 82Unskilled 186 79 0.0005∗∗ 0.001∗∗ 0.001∗∗

Interior platform angle (�)Skilled 422 100Unskilled 186 102 0.015∗∗ 0.001∗∗ 0.0005∗∗

Length/breadthSkilled 453 1.0Unskilled 189 0.9 0.017∗∗ 0.130 0.278

Length/thicknessSkilled 453 7.2Unskilled 189 6.3 0.001∗∗ 0.0005∗∗ 0.0005∗∗

Breadth/thicknessSkilled 453 7.8Unskilled 189 7.3 0.031∗∗ 0.010∗∗ 0.011∗∗

Platform breadth/platformdepthSkilled 438 8.2Unskilled 186 5.6 0.004∗∗ 0.017∗∗ 0.012∗∗

∗∗ Significant at the 95% level.

whole. For this reason, adze-making skill may best beunderstood as a series of production constellations (Kel-ler and Keller 1996) with diverse material, technical, ec-onomic, social, and symbolic dimensions. The exampleprovided by Langda contributes to our understanding ofthe nature of modern human technology and, in so doing,helps to provide a foundation both for the interpretationof archaeological remains and for the reconstruction ofhuman cognitive evolution.

implications for the archaeology of humanorigins

How may these findings regarding the nature of knappingskill in Langda be applied in evaluating the skill of pre-historic knappers? Obviously, the adze-making craft ofLangda is not an exact analogue for any particular pre-historic industry or assemblage. The specific combina-

tion of production goals, techniques, and raw materialsused in Langda are particular to one time and place. De-spite this, the craft does display areas of overlap witha number of different prehistoric technologies (Toth,Clark, and Ligabue 1992).

In the most general terms, the adze industry of Langdaprovides an example of a skill-intensive flaked-stonetechnology designed to produce well-shaped and exten-sively worked large core forms (which in this case aresubsequently ground). Archaeological examples fallingwithin this very broad category include the later Acheu-lean handaxes and cleavers known from sites like Ka-lambo Falls, Zambia (Clark 1969, 1974), Bodo, Ethiopia(Schick and Clark 2000), Lion Spring, Jordan (Copeland1991), St. Acheul, France (Howell 1966), and Boxgrove,England (Bergman and Roberts 1988), as well as MiddleStone Age core axes like those from the Central African“Lupemban” industry and Mesolithic and Neolithic


Fig. 16. Two blade-flakes (utina) produced by askilled knapper during adze-bit shaping.

celts in which the basic form was achieved throughknapping rather than grinding. Obviously, this listincludes a huge amount of technological variation.Within the later Acheulean of Africa alone, a great di-versity of raw-material types, sizes, and shapes were ex-ploited using a range of techniques including the Kom-bewa, “proto-Levallois,” and Talabalat/Tachengit meth-ods (Clark 1994).

Despite this diversity, the unifying goal of producingregular core forms from inherently variable blanks ornodules does lead to certain general convergences. Forexample, it is a simple matter of geometry that thick,short flake removals will reduce only the immediateedge being worked, steepening edges and leaving the in-

terior of a face untouched. Flakes must extend at leasthalf the distance between working edges if uniform re-duction of a face is desired (Callahan 1979), and this basicconstraint applies whether the desired core form is abifacial handaxe or a trihedral adze.

In discussing the manufacture of Acheulean-style bi-faces, Schick (1994:584) lists some common errors madeby beginning knappers, including (1) excessive narrowingof the piece without sufficient thinning, (2) failure tomaintain the piece in a flat plane, (3) production of anuneven outline shape, and (4) failure to impose a bifacialedge on more obtuse areas of the blank. Although theseobservations are made with respect to a particular tech-nology, they illustrate general difficulties in the produc-tion of any well-shaped core form. Difficulties with thin-ning, for example, are clearly related to the problem ofstriking sufficiently invasive flakes discussed above. Cal-lahan (1979:35) describes thinning as a major stumblingblock in the replication of both Acheulean- and Clovis-style bifaces.

Control over outline shape, though partly a product ofhigher-level planning, also relies upon the more basicability to remove trimming flakes in such a way as toavoid gouging and overcorrection. This requires the pro-duction of relatively thin, spreading flakes such as thetypical biface trimming flakes of the Acheulean (Schick1994). In general, the striking of “large, relatively flatand thin flakes with small bulbs of percussion” is es-sential in order to produce large, extensively workedtools with straight edges (Whittaker 1994:185). This isbecause thicker flakes leave deeper flake scars with morepronounced negative bulbs of percussion and thus tendto produce wavy and irregular edges.

To the extent that such general constraints and re-quirements apply across diverse technologies, lessonslearned from the knappers of Langda may be (cautiously)applied to the evaluation of archaeological material. Inany such evaluation, careful analysis of the archaeolog-ical evidence itself, ideally including experimental rep-lication with authentic raw materials, will be a prereq-uisite. Ethnographic analogy can provide importantinsights of a general nature, but application of these toany specific case is appropriate only with referenceto idiosyncratic technological and environmental con-ditions.

Before invoking knapping skill as an explanation forthe observed characteristics of stone artifacts, more par-simonious alternatives should always be considered.Even in the current ethnographic case, for example, somedifferences in flake size and shape were related to dif-ferences in core size. At the same time, when analysisand replication indicate that tool characteristics are un-derdetermined by constraint and least-effort explana-tions, we should not be afraid to credit their makers withthe requisite level of knapping skill.

With this in mind, the following general guidelines forthe evaluation of knapping skill in the archaeologicalrecord, relative to local conditions, may be proposed:

1. Insofar as variation in raw materials may be con-trolled, it requires more skill to detach increasingly large


but relatively thin flakes. One way to produce flakes thatmaximize area but minimize thickness is to exploit plat-form angles approaching 90�, a strategy which requiresthe combination of increased knapping force with pre-served striking accuracy. Strategic skill in the manipu-lation of core morphology, although more difficult toquantify, also plays an important role in the productionof appropriately shaped flakes (e.g., the utina of Langda).Removal of large relatively thin flakes is essential to theproduction of well-formed and extensively worked coreforms and is reflected in the smooth linearity, as opposedto a sinuous “scalloping,” of the edges produced.

2. Where artifact shape is not determined by raw-ma-terial form or reduction intensity, the consistent pro-duction of uniform shapes from nonuniform raw mate-rials reflects the exercise of knapping skills. In Langda,for example, expert knappers display much less varia-bility than novices in the relative proportions of theadzes they produce. Relative to the particular technologybeing considered, more highly standardized artifacts pro-vide evidence of greater levels of skill.

3. Consideration of the evidence from Langda and else-where shows that certain production goals are particu-larly difficult to achieve. As a general rule, it is difficultto produce core forms that maximize one dimension rel-ative to another. Thus, in the Langda material, skilledknappers may be distinguished by their ability to produceadzes that are greatly elongated relative to width andthickness. Similarly, skill in the manufacture of Acheu-lean-style bifaces is reflected in the ability to producepieces that are broad but thin.

4. Finally, raw-material procurement in Langda is ahighly demanding aspect of production involving quitea bit of expertise. Whenever possible, consideration ofraw-material availability should be part of the interpre-tation of archaeological occurrences. In particular, weshould ask how selective the toolmakers were and whatproblems they may have faced in locating, identifying,and exploiting desired materials.

As long as due attention is paid to local conditions andalternative explanations, it is possible to make soundinferences about the relative skill of prehistoric stoneknappers. Such inferences are particularly valuable be-cause the manifestation of skill is intimately related toa wide array of cognitive, social, and economic consid-erations. In Langda, we see an example of a sophisticatedand skill-intensive lithic technology. Even in the struc-tured and nurturing learning environment provided bythe community of craftsmen, skill acquisition can takeyears of diligent effort. Learning and production are con-ditioned by linguistic terminology, social relations, sym-bolic meaning, and metaphor that situate technical per-formance in a humanly meaningful context. Particularlyimportant are the dynamics of task definition and mo-tivation and the way in which they shape task-relatedcognition.

Although the adze industry of Langda is just a singleexample, it does provide a useful source of questions tobe asked about prehistoric chipped-stone technologies.For example, how much time and effort were required

to develop the necessary knapping skills, and what mightthis have meant in social and economic terms? Similarly,what might the distribution and availability of raw ma-terials have meant for relations within and betweengroups? Approaching the issue of cognition more di-rectly, how were knapping tasks defined and motivated?The enormous potential role of social and symbolicmeaning in the acquisition and expression of sophisti-cated knapping skills is underlined by the Langda ex-ample but remains to be demonstrated in other specificcases. Consideration of this issue opens further inter-esting questions regarding the hypothetical cognitive ca-pacities/propensities that might support a comparablelevel of skill in the absence of a comparable social andsymbolic scaffold.

implications for human evolution

All of this is particularly interesting when considered inthe broader context of human cognitive and brain evo-lution. The Langda example is valuable not as a directanalogue for premodern behavior but as a model for theexploration of general issues surrounding the relation-ship between cognition and lithic technology. Such ex-ploration is an essential step toward the generation ofproductive hypotheses regarding the role of stone tech-nology in human cognitive evolution.

Researchers interested in the relationship between pre-historic stone tools and cognition have often emphasizedthe imposition of arbitrary form (Holloway 1969) duringtool making and the shared mental or procedural tem-plates that this is thought to imply (Gowlett 1996). Inthe modern adze industry of Langda, shared ideas andsophisticated planning are essential aspects of produc-tion. The invariant proportions of adze heads producedby skilled knappers clearly reflect the imposition of an“appropriate” (Wynn 1995) form, while analysis of actualoperational sequences illustrates the planning involvedin the concrete realization of this form. Just as important,the impressive perceptual-motor skill displayed byexpert knappers in Langda calls attention to this oft-neglected aspect of human mental uniqueness. It is ap-propriate to ask to what extent these characteristics, ev-ident in at least one modern example, are present inparticular prehistoric technologies.

Prehistoric tool-making skill must also be situated inits broader social context. As observed by Holloway(1981:288), stone tools represent “clues to the more com-plex social processes that were under direct selectionpressures.” Human technology is distinguished by itscomplex interdependence with social structure (K. Gib-son 1993), an evolved condition that should figure prom-inently in any exploration of human uniqueness. Evi-dence of acquired skill is one of the more powerful cluesto “complex social processes” that can be provided bythe study of stone tools.

Research in Langda reveals that the acquisition of so-phisticated knapping skills is far from trivial, even formotivated modern humans with an elaborate system ofsocial support. To the extent that various premodern


hominids might have possessed less elaborate mentalabilities and/or social support, the acquisition of knap-ping skills would have been a proportionately greaterchallenge. Of course, Paleolithic stone technologies likethe Oldowan and earlier Acheulean are notable for theirsimplicity and ease of manufacture. The refined laterAcheulean core forms that emerge ca. 500,000 b.p., how-ever, represent more demanding technologies that maybe replicated by modern researchers only after lengthypractice (e.g., Callahan 1979:35). The presence of suchwell-developed skills in prehistory raises questions aboutsocioeconomic context and attendant mental sophisti-cation, even if the toolmakers were not fully modernHomo sapiens.

The advent or elaboration of mechanisms for the socialfacilitation of skill acquisition clearly must have rep-resented a watershed in the course of hominid evolution,and it may well be that a good part of that which isunique about the human species relates to our relianceupon complex, socially facilitated skills. Kaplan et al.(2000) have constructed such an argument around thecomplexities of modern human hunting skills. To thismay now be added the evidence presented here regardingthe acquisition and performance of modern humanstone-knapping skill.

Conclusion

Research in Langda has provided an example of a modernhuman lithic technology that is remarkable for its so-phistication and complexity. The manufacture of adzeheads by the craftsmen of Langda requires well-devel-oped skills ranging from basic motor coordination andmovement accuracy to higher-level strategic planningand conceptualization. Acquisition of these skills is alengthy and labor-intensive process that is inextricablylinked with the broader social and economic context inwhich adze manufacture takes place.

In Langda, the manufacture of adzes is a social phe-nomenon. Task definition and motivation are providedby participation in the community of craftsmen and bypersonal relationships with trading partners, resourceowners, mythic figures, and the personified stones them-selves. The community also acts as a social scaffold forthe acquisition of skill by apprentices. Structuring rulesand norms, as well as instruction and assistance, providea framework for the arduous process of skill learning.

These findings suggest important questions to beasked about prehistoric stone technologies. Chief amongthese are questions regarding the time and effort requiredfor skill acquisition and the potential social implicationsof these requirements. Attention is also called to themental demands of raw-material procurement.

The broad theoretical issues raised by research inLangda underline the importance of identifying skilledperformance in the archaeological record. Although thistask is greatly complicated by raw-material variation anddifficulties in divining the intent of prehistoric tool-makers, the analysis of lithic products from Langda has

provided some useful insights. These include an appre-ciation of the relative difficulty of producing core formsthat maximize one dimension relative to others and ofstriking large, thin flakes. The careful adaptation of suchgeneral principles to specific archaeological examplesshould provide a means to approach some of the broaderquestions raised in this study.

The observed complexities of adze production inLangda add to our understanding and appreciation of hu-man uniqueness. The complex interdependence of socialstructure and technology and enhanced capacities for thesocial facilitation of skill acquisition in general are de-fining components of the human adaptation. Stone toolsprovide hard evidence of skilled performance that willbe invaluable in determining when and how these ca-pacities emerged.

Comments

blandine bril and valentine rouxEquipe de Recherche Apprentissage et Contexte,EHESS/INSERM 483, 54 Bd. Raspail, 75006 Paris,France ([email protected])/Prehistoire etTechnologie, CNRS/Paris X, 21 allee de l’Universite,92023 Nanterre, France ([email protected]).26 vii 02

As Stout says, “traditional social contexts in which tostudy lithic technologies are very rare in the modernworld.” In fact, it is only in New Guinea and India thatstone knapping is still performed on a sufficiently largescale for the systematic study of different levels of ex-pertise and of the learning process. A group of Frencharcheologists and motion researchers has conductedfieldwork in Khambhat (India) aimed at characterizingthe skills involved in stone knapping (Roux and Pelegrin1988–89, 1989; Roux, Bril, and Dietrich 1995; Bril, Roux,and Dietrich 2000). In Khambhat, stone knapping em-ploys a different technique (indirect percussion) from theone described by Stout among the Langda, but the resultsof the two studies show remarkable convergence.

Both studies compare skilled and unskilled knapperson the basis of the finished products (in New Guineaadzes, in Khambhat beads) and the actions involved intheir manufacture. Both in New Guinea and in India, thefinished products distinguish the two groups; the lessexpert craftsmen produce shorter and smaller piecesoverall. In terms of strategic skills, Langda unskilledcraftsmen pay almost no attention to the dorsal ridge ofthe core, according to Stout because of the basic physicaldifficulty of striking proper invasive flakes. A similartechnical situation in Khambhat leads to similar behav-ior on the part of nonexperts. The ridges of quadrangularrough-outs must be shaped to remove long flakes, andthis shaping requires the production of transversal in-vasive flakes. Less skilled craftsmen pay no attention tothis operation. This suggests that in both cases the less


skilled lack the control of the elementary action (that is,positioning the stone and directing the blow) that is re-quired for mastery of strategic skills.

Apprenticeship also shows strong similarities betweenLangda and Indian craftsmen. It is a progressive process:learners begin with small pieces and only gradually moveon to work on larger ones. It can take ten years to developthe skill needed to produce the largest pieces, whetheradzes or beads. This seems to be a general feature ofexpertise. Another example may be found in the workof Roux and Corbetta (1989, 1990), who carried out afield experiment with potters in India. Here again, fullmastery of the wheel-throwing technique takes ten yearsto acquire. More generally, a period of ten years of ap-prenticeship seems to represent a rule of thumb in othercategories of motor and/or cognitive activity (Ericksonand Lehman 1996).

Stout refers to a perception-action perspective to ex-plain the differences between skilled and unskilledcraftsmen. He concludes from the characteristics offlakes and finished products that these differences prob-ably reflect “the greater ability of experts to exploit andmanipulate core morphology.” From a perception-actionperspective, the demonstration of such a propositionwould require an analysis not only of the finished prod-ucts and overall strategies but also of the elementaryaction. In the Khambhat study, recording of the hammeracceleration by means of an accelerometer showed thatexperts displayed a finer adaptation of the accelerationto the size of the flake. This result is consistent with aperspective that emphasizes the necessary fine-tuning ofthe perception-action cycle. We concluded in our 1995paper that “what must be learned is not the movementper se, which is quite simple, but the optimal adaptationto the constraints of the environment” (here, the mor-phological and geometrical properties of the stone andthe weight, dimensions, and material of the hammer).

Stout raises the question of the relevance of the Langdacase to the study of prehistoric knapping skills. However,uncovering pieces of the puzzle will require the consid-eration and comparison of existing ethno-archeologicaland experimental studies. Only in this way will it bepossible to highlight the invariant underlying mecha-nisms at work in complex situated action in general andknapping in particular.

sophie a . de beauneUniversite Jean Moulin-Lyon III, UMR CNRS 7041“Archeologies et Sciences de l’Antiquite,” 2 ruePeclet, F-75105 Paris, France ([email protected]). 25 vi 02

Stout’s work in an Irian Jayan village provides a greatdeal of information on the technical competence devel-oped by artisans who make stone adzes. It shows thatthis competence is not simply a matter of motor andcognitive conditions but also depends heavily on the ec-onomic and social context, from the selection of the rawmaterial to the hafting of the shaped and polished adze

head. However, the morphometric study of these adzeheads, interesting as it is, is somewhat disappointing inthat the results obtained only confirm what intuitionhas already suggested: the dimensions of the adze heads,their proportions, their regularities, their sophisticationare associated with the degree of skill of the artisan, themost accomplished adze heads being the work of theexperts as opposed to the apprentices. The requirementsfor developing these abilities are many, and I will confinemyself here to those that may have important archaeo-logical implications.

The period of apprenticeship required to acquire thesestone-working skills is very long (minimum five years),and the richness of the linguistic and technical ex-changes between the skilled and the unskilled (discus-sion, observation, demonstration, and even direct assis-tance) is remarkable. Talking about actions is, however,far from universal. Cases of apprenticeship by imitationalone are legion, judging from the ethnographic evidence,and do not affect the richness of the technical vocabulary,especially in oral-tradition societies: vocabulary can berich and precise without being employed in apprentice-ship as such. It serves less to teach than to comment,on occasion, on what is being taught by imitation.

The archaeological implications suggested by Stout areinteresting. He suggests that the production of bifacesimplies the possession of the social and mental struc-tures necessary for apprenticeship and the production ofsuch tools since the terminal Acheulean (500,000 yearsago). If one agrees with him that it is the interactionamong tool kit, language, and society that would haveproduced the skills necessary for the production of suchstone adzes, this would mean that the acquisition of lan-guage took place very much earlier than anthropologistshave supposed—as early as the first bifaces. This possi-bility can be considered, but two objections to it can beraised. First, we have seen that the transmission ofknowledge does not require language. Furthermore, along period of apprenticeship for the acquisition of tech-nical skill is not unique to humans. In groups of chim-panzees that crack nuts with hammer and anvil, youngchimpanzees spend several years learning how to do it.The kind of apprenticeship is of course not the same (themother chimpanzee rarely guides her infant and has norecourse to any verbal explication), and the motor prob-lems involved are different, but the fact remains thatanimal socialization suffices to create the conditions fora long apprenticeship. We must therefore be cautiouswith regard to the kind of socialization that the produc-tion of bifaces implies for the Acheuleans.

Stout also suggests a new way of approaching the ques-tion of the origin of flint knapping and of hominizationin his attempt to determine at what point in evolutionthe motor, cognitive, social, and economic conditions forthe emergence of stone-working skills came together.Although his argument is appealing, it is preliminary andraises a number of questions. I look forward to learninghow Stout envisages testing his model with regard to thefirst hominid stone workers and their skills.


j . a . j . gowlettDepartment of Archaeology, University of Liverpool,Liverpool L69 3GS, U.K. ([email protected]). 30 vii 02

The analysis of Langda technology presented by Stoutforms a very valuable case study, especially in its trac-ing of social context and careful documentation oftechnical features and their significance at the indi-vidual level. Each such study is precious. Those avail-able already show great variety in the factors inter-acting to shape technological production, as Stoutnotes—from the entirely practical approach of theGalla, whose scrapers must be shaped to fit theirwooden handles (Gallaher 1977), to the example givenby Paton (1994) for northern Australia, where the en-tire purpose of leilira blade production appears to besymbolic and, presumably, functional constraintswould no longer apply.

As Stout observes, it is therefore difficult to trian-gulate from modern evidence to premodern humans,but in comparisons there is some scope to considerthe social embeddedness of chimpanzee technology(McGrew 1992) and also to analyse consistencies andinconsistencies in the production of ancient assem-blages. As the tree of hominid origins becomes bush-ier, it becomes a question of choice whether our majorpurpose is to clarify the distinctiveness of modern hu-manity or to make some effort to map behavior ontothe various past hominid representatives.

Stout clearly traces the range of factors contributingto the final form of a tool, giving an explanation ofsize variation that may be specific to the Langda andperhaps related peoples. Size variation has been ex-plored for the Acheulean, but with little explanationfor its social or functional context. The various factorsof variation make the word “template,” which manyof us use, seem too hard or immutable. One can at-tempt alternatives such as “schema” or “instructionset,” but the point emerges that we have very littleknowledge of how much factors such as raw material,functional need, or the standing of the particular knap-per affect the form of the output (in a way which mighteven vary from day to day). Even so we can see fromthe Acheulean—where, paradoxically, we have largercomparative datasets than for the present, though dif-ficulty in unravelling the message—fairly regularlength variation in bifaces from about 8 to 25 cm. Whydoes this recur over large areas and through greatdepths of time? I have always been struck that onesees this range even if one finds a only dozen bifacesin a landscape. Studies such as Stout’s provide someinsights into possibilities and suggest approaches thatwill allow us to aim to see, say, Acheulean assem-blages as the products of communities of individualsand attempt to disentangle some of the factors bearingupon them.

charles m. kellerDepartment of Anthropology, University of Illinois,Urbana, Ill. 61801, U.S.A. ([email protected]). 31 vii 02

Stout usefully applies the Vygotskian notion of thethought/action dialectic in his attempt “to develop pro-ductive hypotheses about the psychological implicationsof prehistoric stone tools.” But to treat the conceptualfacet of this dynamic interrelationship simply as “think-ing” fails to take advantage of the information that hisotherwise insightful description provides. He presentsevidence for a considerable range of cognitive processesin the production of the stone adzes in question.

Reflection, in the form of recalling past actions andconsidering options for future operations, takes placeduring both periods of active knapping and intervals ofinactivity. Identification of various stages in the produc-tion process and construction of a shared terminologyare indications of reflection by individuals and collectivecommunication among the adze makers. Reflection onthe success or failure of previous knapping episodes alsoguides the master stone worker in his selection and al-location of boulders for flaking.

Integration of diverse technological and social infor-mation takes place throughout the manufacturing pro-cess as personal and group relations, social norms, andmythic significance influence the attitudes and proce-dures employed. The scaffolding provided by the com-munity of practice allows the construction of expertiseby the apprentices and provides resources for the moreadept when problematic conditions arise. The situatednature of the activity adds other practitioners to theideas, tools, and materials that make up the productionconstellations employed.

Monitoring the progress of each piece requires con-stant comparison of mental images of the goal form, aswell as appropriate intermediate shapes, with the actualstate of the work piece. The fact that the products ofexperts differ in their proportions from those of the ap-prentices is an indication of the critical role of mentalimagery in guiding motor behavior. Skill is grounded asmuch in sophisticated imagery as in manual dexterity,and the complexity of its construction contributes to thelength of the apprenticeship that novices serve.

Planning takes place before the desired sequence ofcore reduction steps is initiated and continues during theoperation. The emergent qualities of the process are re-vealed as each flake removal is evaluated and subsequentactions are considered. Failure of a planned removal re-sults in the application of corrective measures derivedfrom a repertoire of solutions based on individual ex-perience and a shared stock of knowledge.

Understanding of ethnographic information or the re-mains of a prehistoric atelier can be enhanced by con-sidering cognition as a set of processes rather than a mon-olithic unit.


thomas wynnDepartment of Anthropology, University of Colorado,Colorado Springs, Colo. 80903, U.S.A. ([email protected]). 8 viii 02

Any ethnographic description of stone knapping has in-trinsic value, if for no other purpose than to check un-derstandings derived from studies of nontraditional stoneknapping. However, as Stout illustrates, ethnographicstudies can also supply an account of the social contextof stone knapping. Such accounts are few and thereforevery valuable both for our understanding of technicalactivity in general and for the archaeology of humanevolution.

My reaction to Stout’s analysis is that it is overly cau-tious and conservative. Stout casts his theoretical netvery widely, invoking the ideas of Vygotsky, Gibson, In-gold, Keller and Keller, and others. Instead of distillingthese ideas into a single coherent account of technicalaction and cognition, he brings them to bear in a piece-meal fashion. As a consequence, it is difficult to appre-ciate how his account of Langda stone knapping enrichesour understanding of stone knapping and technical ac-tion in general. This is, of course, the point. What canStout now tell us about the cognitive context of stoneknapping (and technique in general) that we did notknow before? Simply referring to the “zone of proximaldevelopment” or “constellations of knowledge” does notquite do this. It is perhaps too much to expect him toconstruct a new grand theory, but even tentative at-tempts to formulate a single and consistent account ofanalytical concepts might have yielded new insights.

Stout is a bit bolder in his discussion of the implica-tions for archaeology and human evolution. He proposesfour guidelines for the assessment of knapping skill, andwhile these may strike some as obvious they acquireanalytical value precisely because they derive from theethnographic evidence (as opposed to the musings of theresearcher or even studies of replication). What I wouldlike to have seen is a bit more venturesome approach tothe archaeological record itself. Stout suggests, for ex-ample, that the advent of “social facilitation of skill ac-quisition” represented an evolutionary watershed. Whendoes he think this occurred? He implies that he favorsa Late-Acheulean time frame (but perhaps my bias colorsmy reading). This would be a very important contribu-tion to the archaeology of human evolution, which isStout’s stated ultimate goal, and I wish he had devotedmore than a few sentences to the hypothesis.

My final point is based on an interpretation of the textthat Stout perhaps did not intend, but as it bears on animportant methodological issue it is worth airing. Hesuggests at the beginning of his text that “our interpre-tation of the lithic evidence must ultimately rely uponmodels developed from actualistic research.” If we takethis to mean that our understanding of the past is de-pendent on our understanding of the present, this state-ment provokes no quarrel. However, if it means thatunderstanding of prehistoric stone knappers can only de-rive from actualistic studies of modern stone knapping

or even that such studies must be granted priority, thenI must disagree. It seems to me that all studies of skilledactivity can provide potential analogues and that moregeneral accounts of human and nonhuman cognition canprovide important insights (as Stout himself has done inciting Vygotsky, Gibson, etc., none of whom knew athing about stone knapping).

Reply

dietrich stoutBloomington, Ind., U.S.A. 7 xii 02

I thank the commentators for the time and effort thatthey have invested in responding to this article. Spaceis a real constraint when dealing with a subject thattouches on such a diverse body of method, theory, andevidence, and I am grateful for the opportunity to elab-orate on particular points that the commentators havefound interesting and/or troublesome.

Both Wynn and de Beaune consider the analysis I havepresented somewhat disappointing, although for differ-ent reasons. For example, whereas de Beaune regrets thatthe analysis of artifacts “only confirm[s] what intuitionhas already suggested,” Wynn points out that these re-sults have value “precisely because they derive from theethnographic evidence” rather than from “the musingsof the researcher.” Here I agree with Wynn. In fact, theprimary goal of my research in Langda was to contributeto a more methodical and empirical approach to the sub-ject of stone tools and cognition. Perhaps this is why,despite our points of agreement, Wynn considers myanalysis overly conservative.

It is difficult to respond to Wynn’s general theoreticalcriticisms without knowing what specific deficiencieshe has identified, so I will settle for simply attemptingto clarify my perspective. Much previous research on thesubject of stone tools and cognition has been guided bya theoretical perspective that I would characterize ascomputational and essentialist. Cognition is viewed asan abstract system of formal computations carried outin the brain, with sensation and action being treated aslittle more than peripheral input/output channels. Thepsychological significance of stone tools is thus to beunderstood in terms of essential cognitive operationssuch as “hierarchical combination” (Greenfield 1991,Matsuzawa 1996) or “spatio-temporal substitution”(Wynn 1989) rather than in terms of overtly variable toolbehaviors. For example, such apparently dissimilar be-haviors as bifacial Oldowan flaking and the stripping ofsedge stems by chimpanzees might be seen as essentiallyequivalent examples of an underlying “property ofboundary” (Wynn and McGrew 1989).

This approach produces important insights on onelevel but could benefit from greater attention to the con-crete details of performance. To this end, I follow Roux,Bril, and Dietrich (1995) in adopting a perception-action


approach to stone knapping. This is not the appropriateplace to review the substantial body of theory and re-search in dynamic systems and ecological psychologythat contributes to the perception-action approach (seeThelen and Smith 1994, Reed and Bril 1996, Smitsman1997, van Gelder 1998, Lockman 2000), but the basicargument is that skilled behaviors such as stone knap-ping emerge through the effortful coordination of per-ception and action rather than simply devolving from anabstract internal system of formal cognitive operations.

Bril and Roux have provided a useful summary of re-search among the stone bead knappers of Khambhat, In-dia, that, in conjunction with my own work in Langda,highlights the fundamental role of perceptual-motor co-ordination in the emergence of knapping skill. In bothcases, optimization of the motor synergies involved ineffective flake removal provides the necessary means forthe realization and stabilization of larger-scale knappingpatterns (strategies). Less skilled craftsmen “know”about the operations needed to produce high-qualitybeads or adze heads but do not really comprehend themin the same experiential way as do more skilled knap-pers. For modern humans at least, it is much more dif-ficult and time-consuming to acquire knapping skill thanit is to develop a theoretical knowledge of the appropriatestages of reduction.

For this reason, consideration of the context and dy-namics of knapping skill acquisition should figure prom-inently in psychological interpretations of stone tools.As Wynn points out, there is nothing new in the conceptsof situated learning and distributed cognition that I haveemployed in explicating the Langda example. Wynn him-self (1995) has drawn on similar ideas, and Tomasello(1999) has written at length about the “cultural originsof human cognition.” The value of my case study inLangda is in providing a concrete example of the rele-vance of a cultural psychological (Bruner 1990) perspec-tive to stone knapping. Of central importance is the ob-servation that, although the behavioral experimentationneeded to achieve knapping skill is an individual activ-ity, it is motivated and enabled by an elaborate systemof cultural meanings and social relations.

In Langda this system includes linguistic communi-cation, but this need not have been the case with pre-modern technologies. I do not agree with de Beaune thatmy argument implies that the acquisition of languagetook place “as early as the first bifaces.” As an anony-mous reviewer of my original manuscript pointed out,Langda provides a prime example of the ostensive use oflanguage “as a means of drawing listeners’ attention intothe world.” This can also be achieved by nonlinguisticmeans. Donald (1991), for example, provides a specula-tive but intriguing account of what a prelinguistic, mi-metic culture might have looked like.

The important point is not the necessity of languagebut the cultural nature of technical learning and practice.Although de Beaune contends that “cases of apprentice-ship by imitation alone are legion” and that “appren-ticeship for the acquisition of technical skill is notunique to humans,” I would favor a more restricted use

of the term “apprenticeship.” For humans, apprentice-ship implies participation in a structured and meaningfulcommunity of practice (Lave and Wenger 1991) that sup-ports and motivates learning. Such communities growout of routinely intersubjective (Quine 1960) social re-lations in which the understanding of others as inten-tional agents creates a medium for shared meaning, at-tention, and action, including true pedagogy.

In contrast, de Beaune notes in her example of chim-panzee nut-cracking that “the mother rarely guides herinfant” (see also Inoue-Nakamura and Matsuzawa 1997).She speaks of teaching “by imitation,” and while it mayseem a quibble to point out that imitation is a means oflearning rather than of teaching (the pedagogical coun-terpart would be demonstration), such distinctions withrespect to intentionality are actually pivotal. Ape soci-eties do provide a certain degree of scaffolding for skillacquisition, but I would contend that this scaffoldingdiffers in critical ways from human apprenticeship, es-pecially with respect to the prevalence of shared mean-ings. Over the course of human evolution, a distinctive,cultural mode of skill acquisition has emerged, and I aminclined to follow Tomasello (1999) in suggesting that ithas its foundations in the saturation of everyday socialinteractions with intersubjective awareness.

For this reason, I feel that a focused and empiricalprogram of research into the archaeological evidence ofhominid knapping skill has much to contribute to thestudy of human cognitive evolution. It would appear thatBril, Roux, and Gowlett share this feeling. In fact, Briland Roux take me to task for failing to provide a kine-matic analysis of the elementary knapping action. I haveargued that the artifact attributes I report do provide in-direct evidence of the elementary action (evidence thatwould also be observable in the archaeological record),but I would certainly like to expand my research in thisdirection in the future.

De Beaune is interested to know how I envisage testingmy ideas regarding skill and social facilitation in earlyhominid stone knappers. Although these ideas representmore of a research orientation than a set of testable hy-potheses, I would advocate focused archaeological casestudies conducted in conjunction with experimental in-vestigations of knapping skill acquisition. I agree withGowlett’s comments regarding the importance of vari-ation in lithic technologies and feel that given the cur-rent state of our knowledge we are more in need of spe-cific examples than of sweeping syntheses.

Wynn is disappointed that I have not been more ven-turesome in my interpretation of the archaeological rec-ord, but I would contend that any substantive new in-terpretations will require further primary research. I haveventured to note the general sophistication of laterAcheulean knapping and to argue that such “well-de-veloped skills” raise questions about “socioeconomiccontext and attendant mental sophistication.” However,answering such questions will necessarily be an exercisein specifics. A perception-action perspective demandsthat the psychological interpretation of stone tools befirmly rooted in the details of knapping behavior, and I


would reemphasize that the detailed evaluation of knap-ping skill can occur only with respect to specific localconditions. Given the analytic value of specificity, Iwould argue, contra Wynn, that research on stone knap-ping per se should indeed be a priority for cognitive ar-chaeologists, although I agree that studies of “any skilledactivity” are valuable.

I am particularly grateful to Keller for discussing theissue of diversity in knapping-related cognitive pro-cesses, an important (and extensive) subject that I havelargely glossed over in this article. Keller speaks of threekinds of cognitive activity: reflection, mental imagery,and planning. With “reflection” he appears to be high-lighting the same distinction between “doing” and“thinking about doing” that I have picked up from thework of Thelen and Smith (1994). Although I wouldstress that reflection and performance are both kinds ofaction, they generally occur on different spatio-temporalscales and probably involve different (overlapping?) neu-ral substrates and dynamics (cf. Green 2001).

Although Keller’s comments reveal the value ofLangda as an example of knapping-related cognitive pro-cesses, controlled experimentation will be important infurther dissecting these processes. For example, Kellermaintains that “skill is grounded as much in sophisti-cated imagery as in manual dexterity.” Preliminary func-tional brain-imaging research (Stout et al. 2000) with Ol-dowan-style knapping, however, shows that brainregions classically associated with mental imagery(Kosslyn, Ganis, and Thompson 2001) are not heavilyrecruited during this activity. Such experimental ap-proaches will represent an important supplement to in-trospection in assessing the role of cognitive processessuch as “imagery” and “planning” in specific knappingtechniques.

References Citeda m b ro s e , s t a n l e y. 2001. Paleolithic technology and human

evolution. Science 291:1748–53.b e r g m a n , c . a . , a n d m . b . ro b e r t s . 1988. The Lower

Paleolithic site at Boxgrove, West Sussex. Revue Archaeolo-gique de Picardie 1–2:105–14.

b e r n s t e i n , n i k o l a i . 1967. The coordination and regulationof movements. New York: Pergamon.

b o r d e s , f r a n c o i s . 1947. Etude comparative des differentestechniques de taille du silex et des roches dures. L’Anthro-pologie 51:1–29.

b r a n d t , s t e v e n . 1996. “The ethnoarchaeology of flakedstone tool use in southern Ethiopia,” in Aspects of African ar-chaeology: Proceedings of the Tenth Congress of the Pan Afri-can Association for Prehistory and Related Studies. Harare.

b r i l , b . , v. ro u x , a n d g . d i e t r i c h . 2000. “Habiletesimpliquees dans la taille des perles en calcedoine: Caracteris-tiques motrices et cognitives d’une action situee complexe,” inCornaline de l’Inde: Des pratiques techniques de Cambay auxtechno-systemes de l’Indus. Edited by V. Roux, pp. 207–332.Paris: Editions de la MSH. [bb, vr]

b ru n e r , j e ro m e . 1990. Acts of meaning. Cambridge: Har-vard University Press.

b u r t o n , j o h n . 1985. Axe makers of the Wahgi: Pre-colonial

industrialists of the Papua New Guinea Highlands. Ph.D. diss.,Australian National University, Canberra, Australia.

c a l l a h a n , e r r e t t . 1979. The basics of biface knapping inthe Eastern Fluted Point Tradition: A manual for flintknappersand lithic analysts. Archaeology of Eastern North America 7:1–172.

c a n e , s c o t t . 1992. Aboriginal perceptions of their stone tooltechnology: A case study from the Western Desert, Australia.Australian Archaeology 35:11–31.

c l a r k , j . d e s m o n d . 1969. Kalambo Falls prehistoric site.Vol. 1. Cambridge: Cambridge University Press.

———. 1974. Kalambo Falls prehistoric site. Vol. 2. Cambridge:Cambridge University Press.

———. 1994. “The Acheulean industrial complex in Africa andelsewhere,” in Integrative paths to the past: Paleoanthropolog-ical advances in honor of F. Clark Howell. Edited by R. S. Cor-ruccini and R. L. Ciochon, pp. 451–69. Englewood Cliffs: Pren-tice-Hall.

c l a r k . j . d e s m o n d , a n d h i ro k u r a s h i n a . 1981. “Astudy of the work of a modern tanner in Ethiopia and its rele-vance for archaeological interpretation,” in Modern materialculture. Edited by R. A. Gould and M. B. Schiffer, pp. 303–21.New York: Academic Press.

c o p e l a n d , l o r r a i n e . 1991. The late Acheulian knapping-floor at C-Spring, Azraq Oasis, Jordan. Levant 23:1–6.

c r a b t r e e , d o n . 1966. A stoneworker’s approach to analyzingand replicating the Lindenmeier Folsom. Tebiwa 9(1):3–39.

d i b b l e , h a ro l d l . , a n d a n d r e w p e l c i n . 1995. Theeffects of hammer mass and velocity on flake mass. Journal ofArchaeological Science 22:429–39.

d o n a l d , m e r l i n . 1991. Origins of the modern mind: Threestages in the evolution of culture and cognition. Cambridge:Harvard University Press.

d r i s k e l l , j . , c . c o p p e r , a n d a . m o r a n . 1994. Doesmental practice enhance performance? Journal of Applied Psy-chology 79:481–92.

e r i c s o n , k . a . , a n d a . c . l e h m a n . 1996. Expert and ex-ceptional performance: Evidence from maximal adaptation totask constraints. Annual Review of Psychology 47:273–305.[bb, vr]

f i t t s , p . m . 1954. The information capacity of the humanmotor system in controlling the amplitude of movement. Jour-nal of Experimental Psychology 47:381–91.

g a l l a g h e r , j a m e s p . 1977. Contemporary stone tools inEthiopia: Implications for archaeology. Journal of Field Archae-ology 4:407–14.

g a t e w o o d , j o h n . 1985. “Actions speak louder than words,”in Directions in cognitive anthropology. Edited by J. Dough-erty, pp. 199–219. Urbana: University of Illinois Press.

g i b s o n , j a m e s j . 1979. The ecological approach to visualperception. Boston: Houghton Mifflin.

g i b s o n , k a t h l e e n r . 1993. “Tool use, language, and socialbehavior in relationship to information processing capacities,”in Stone tools, language, and cognition in human evolution.Edited by T. Ingold and K. Gibson, pp. 251–70. Cambridge:Cambridge University Press.

g i b s o n , k a t h l e e n r . , a n d t i m i n g o l d . Editors. 1993.Stone tools, language, and cognition in human evolution.Cambridge: Cambridge University Press.

g o u l d , r i c h a r d a . 1980. Living archaeology. Cambridge:Cambridge University Press.

g o w l e t t , j o h n a . j . 1996. “Mental abilities of early Homo:Elements of constraint and choice in rules systems,” in Model-ing the early human mind. Edited by P. Mellars and K. R. Gib-son, pp. 191–216. Exeter: Short Run Press.

g r a v e s , b e n n i e . 1989. “Informal aspects of apprenticeship inselected American occupations,” in Apprenticeship: From the-ory to method and back again. Edited by M. W. Coy, pp.51–64. Albany: State University of New York Press.

g r a v e s , p a u l . 1994. Flakes and ladders: What the archaeolog-ical record cannot tell us about the origins of language. WorldArchaeology 26:158–71.

g r e e n , p a t r i c k r . 2001. The relation between perception


and action: What should neuroscience learn from psychology?Ecological Psychology 13:117–22.

g r e e n fi e l d , p a t r i c i a m . 1991. Language, tools, and brain:The development and evolution of hierarchically organized se-quential behavior. Behavioral and Brain Sciences 14:531–95.

h a m p t o n , o . w. 1999. Culture of stone: Sacred and profaneuses of stone among the Dani. College Station: Texas A&MUniversity Press.

h a y d e n , b r i a n . 1977. “Stone tool functions in the WesternDesert,” in Stone tools as cultural markers: Change, evolu-tion, and complexity. Edited by R. V. S. Wright, pp. 178–88.Canberra: Australian Institute of Aboriginal Studies.

h o d d e r , i a n . 1990. Technology in the humanities: A com-mentary. Archaeological Review from Cambridge 9:154–57.

h o l l o w a y, r a l p h l . 1969. Culture, a human domain. cur-rent anthropology 10:395–412.

———. 1981. Culture, symbols, and human brain evolution: Asynthesis. Dialectical Anthropology 5:287–303.

h o w e l l , f . c l a r k . 1966. Observations on the earlier phasesof the European Lower Paleolithic. American Anthropologist68(2), pt. 2:88–201.

i n g o l d , t i m . 1993. “Tool use, sociality, and intelligence,” inStone tools, language, and cognition in human evolution. Ed-ited by T. Ingold and K. Gibson, pp. 429–46. Cambridge: Cam-bridge University Press.

———. 1997. Eight themes in the anthropology of technology.Social Analysis 41:106–38.

i n o u e - n a k a m u r a , n o r i k o , a n d t e t s u ro m a t s u -z a w a . 1997. Development of stone tool use by wild chimpan-zees (Pan troglodytes). Journal of Comparative Psychology 111:159–73.

i s a a c , g l y n n . 1986. “Foundation stones: Early artifacts asindicators of activities and abilities,” in Stone Age prehistory.Edited by A. Berthelet and J. Chavaillon, pp. 221–41. Cam-bridge: Cambridge University Press.

k a p l a n , h . , k . h i l l , j . b . l a n c a s t e r , a n d a . m .h u r t a d o . 2000. A theory of human life history evolution:Diet, intelligence, and longevity. Evolutionary Anthropology 9:156–85.

k e i j z e r , f r e d a . 1999. Doing without representations thatspecify what to do. Philosophical Psychology 11:269–302.

k e l l e r , c h a r l e s m . , a n d j a n e t d . k e l l e r . 1996.Cognition and tool use: The blacksmith at work. Cambridge:Cambridge University Press.

k o s s l y n , s . m . , g . g a n i s , a n d w. l . t h o m p s o n .2001. Neural foundations of imagery. Nature Reviews: Neuros-cience 2:635–42.

l a v e , j e a n , m . m u r t a u g h , a n d o . d e l a ro c h a .1984. “The dialectic of arithmetic in grocery shopping,” inEveryday cognition: Its development in social context. pp.67–94. Cambridge: Harvard University Press.

l a v e , j . , a n d e . w e n g e r . 1991. Situated learning: Legiti-mate peripheral participation. Cambridge: Cambridge Univer-sity Press.

l e ro i - g o u r h a n , a n d r e . 1965. Le geste et la parole 2: Lamemoire et les rhythmes. Paris: A. Michel.

l i g a b u e , g i a n c a r l o . 1985. The stone axes of the pygmiesof New Guinea. Ligabue Magazine 4(6):30–35.

l o c k m a n , j e f f r e y j . 2000. A perception-action perspectiveon tool use development. Child Development 71:137–44.

l o u w e r s e , j . 1990. “A traditional causality and adaptation toenvironment: Repercussion of cognition of environment on so-ciocultural interaction,” in The Mek and their neighbors (In-ternational Conference, Max Planck Institute in Seewiesen).Seewiesen.

m c g r e w, w. r . 1992. Chimpanzee material culture: Impli-cations for human evolution. Cambridge: Cambridge Univer-sity Press. [jajg]

m a r z k e , m a ry, n . t o t h , k . s c h i c k , s . r e e c e , b .s t e i n b e r g , k . h u n t , r . l . l i n s h e i d , a n d k . - n .a n . 1998. EMG study of hand muscle recruitment during hardhammer percussion manufacture of Oldowan tools. AmericanJournal of Physical Anthropology 105:315–32.

m a t s u z a w a , t e t s u ro . 1996. “Chimpanzee intelligence innature and in captivity: Isomorphism of symbol use and tooluse,” in Great ape societies. Edited by W. McGrew, L. Mar-chant, and T. Nishida, pp. 196–209. Cambridge: CambridgeUniversity Press.

n e w e l l , k a r l m . 1996. “Change in movement and skill:Learning, retention, and transfer,” in Dexterity and its devel-opment. Edited by M. L. Latash and M. T. Turvey, pp. 393–430.Mahwah, N.J.: Lawrence Erlbaum Associates.

o a k l e y, k e n n e t h p . 1954. “Skill as a human possession,”in A history of technology, vol. 1, From earliest times to thefall of ancient empires. Edited by C. Singer, E. J. Holmyard,and A. R. Hall, pp. 1–37. Oxford: Oxford University Press.

p a r k e r , s u e t a y l o r , a n d k a t h l e e n g i b s o n . 1979. Amodel of the evolution of language and intelligence in earlyhominids. Behavioral and Brain Sciences 2:367–407.

p a t o n , r . 1994. Speaking through stones: A study from north-ern Australia. World Archaeology 26:172–84. [jajg]

p e l e g r i n , j a c q u e s . 1993. “A framework for analyzing pre-historic stone tool manufacture and a tentative application tosome early stone industries,” in The use of tools by humanand non-human primates. Edited by A. Berthelet and J. Cha-vaillon, pp. 302–14. Oxford: Clarendon Press.

p e t r e q u i n , p i e r r e , a n d a n n e - m a r i e p e t r e q u i n .1993. Ecologie d’un outil: La hache de pierre en Irian Jaya (In-donesie). Paris: CNRS Editions.

p r e s t o n , b e t h . 1998. Cognition and tool-use. Mind and Lan-guage 13:513–47.

q u i n e , w. v. 1960. Word and object. Cambridge: HarvardUniversity Press.

r e e d , e d w a r d s . , a n d b l a n d i n e b r i l . 1996. “The pri-macy of action in development,” in Dexterity and its develop-ment. Edited by M. L. Latash and M. T. Turvey, pp. 431–52.Mahwah, N.J.: Lawrence Erlbaum.

ro u x , v a l e n t i n e . 1990. The psychological analysis of tech-nical activities: A contribution to the study of craft specializa-tion. Archaeological Review from Cambridge 9:142–53.

ro u x , v a l e n t i n e , b . b r i l , a n d g . d i e t r i c h . 1995.Skills and learning difficulties involved in stone knapping.World Archaeology 27:63–87.

ro u x , v. , i n c o l l a b o r a t i o n w i t h d . c o r b e t t a ,1989. The potter’s wheel: Craft specialization and technicalcompetence. New Delhi: Oxford and IBH Publishing.

———. 1990. Le tour du potier: Specialisation artisanale et com-petences techniques. Paris: Editions du CNRS. [bb, vr]

ro u x , v. , a n d j . p e l e g r i n . 1988–89. Knapping techniqueand craft specialization: An ethnoarchaeological investigationin Gujarat. Puratattva 19:50–59.

———. 1989. Taille des perles et specialisation artisanale: En-quete ethnoarcheologique dans le Gujarat. Techniques et Cul-ture 14:23–49. [bb, vr]

s a l v a t o r i , s a n d ro . 1986. “La asce di petra dei Papua pig-mei,” in Indonesia: La grande deriva etnica. Edited by G. M.Ronzoni, pp. 147–52. Venice: Erizzo Editrice.

s c h i c k , k a t h y d . 1994. “The Movius line reconsidered: Per-spectives on the earlier Paleolithic of Eastern Asia,” in Integra-tive paths to the past: Paleoanthropological advances in honorof F. Clark Howell. Edited by R. S. Corruccini and R. L. Cio-chon, pp. 569–95. Englewood Cliffs: Prentice-Hall.

s c h i c k , k a t h y d . , a n d j . d e s m o n d c l a r k . 2000.“Acheulean archaeology of the Western Middle Awash,” inThe Acheulean and Plio-Pleistocene deposits of the MiddleAwash Valley, Ethiopia. Edited by J. de Heinzelin, J. D. Clark,K. D. Schick, and W. H. Gilbert, pp. 123–82. (Annals of Geo-logical Science 104.) Tervuren: Royal Museum of CentralAfrica.

s c h i f f e r , m i c h a e l . 1999. The material life of human be-ings: Artifacts, behavior, and communication. New York:Routledge.

s k e r t c h l y, s y d n e y b . j . 1879. On the manufacture ofgun-flints, the methods of excavating for flint, the age of Pale-olithic man, and the connexion between Neolithic art and the


gun-flint trade. London: Geological Survey of England andWales.

s m i t s m a n , a . w. 1997. “The development of tool use:Changing boundaries between the organism and environment,”in Evolving explanations of development: Ecological ap-proaches to organism-environment systems. Edited by C.Dent-Read and P. Zukow-Goldring. Washington, D.C.: Ameri-can Psychological Association.

s t e r n s b e r g , r . j . , r . k . w a g n e r , a n d l . o k a g a k i .1993. “Practical intelligence: The nature and role of tacitknowledge in work and school,” in Mechanisms of everydaycognition. Edited by J. Puckett and H. Reese, pp. 205–27. Hills-dale, N.J.: Lawrence Erlbaum Associates.

s t o u t , d . , n . t o t h , k . s c h i c k , j . s t o u t , a n d g .h u t c h i n s . 2000. Stone tool-making and brain activation:Positron emission tomography (PET) studies. Journal of Ar-chaeological Science 27:1215–23.

s t r a t h e r n , m a r i l y n . 1970. Stone axes and flake tools:Evaluations from New Guinea. Proceedings of the PrehistoricSociety, n.s., 35:311–29.

t h e l e n , e s t h e r , a n d l i n d a s m i t h . 1994. A dynamicsystems approach to the development of cognition and action.Cambridge: MIT Press.

t i n d a l e , n . b . 1965. Stone implement making among theNkako, Ngadadjara, and Pitjandjara of the Great Western Des-ert. Research of the South Australian Museum 15:131–64.

t o m a s e l l o , m i c h a e l . 1999. The cultural origins of humancognition. Cambridge: Harvard University Press.

t o t h , n i c h o l a s . 1985. The Oldowan reassessed: A close

look at early stone artifacts. Journal of Archaeological Science12:101–20.

t o t h , n i c h o l a s , j . d e s m o n d c l a r k , a n d g i a n -c a r l o l i g a b u e . 1992. The last stone axe makers. Scien-tific American 267(1):88–93.

v a n g e l d e r , t i m . 1998. The dynamical hypothesis in cogni-tive science. Behavioral and Brain Sciences 21:615–65.

v i a l , l . g . 1940. Stone axes of Mt. Hagen, New Guinea. Oce-ania 11:158–63.

v y g o t s k y, l e v s . 1976. Mind in society: The development ofhigher psychological processes. Cambridge: Harvard UniversityPress.

w e e d m a n , k a t h ry n . 2000. An ethnoarchaeological study ofstone scrapers among the Gamo people of southern Ethiopia.Ph.D. diss., University of Florida, Gainesville, Fla.

w h i t t a k e r , j o h n c . 1994. Flintknapping: Making and un-derstanding stone tools. Austin: University of Texas Press.

w o o d , d . , j . s . b ru n e r , a n d g . ro s s . 1976. The role oftutoring in problem solving. Journal of Child Psychology andPsychiatry 17:89–100.

w y n n , t h o m a s . 1989. The evolution of spatial competence.Urbana: University of Illinois Press.

———. 1993. “Layers of thinking in tool behavior,” in Stonetools, language, and cognition in human evolution. Edited byT. Ingold and K. Gibson, pp. 389–406. Cambridge: CambridgeUniversity Press.

———. 1995. Handaxe enigmas. World Archaeology 27:10–24.w y n n , t h o m a s , a n d w i l l i a m m c g r e w. 1989. An

ape’s view of the Oldowan. Man 24:383–98.

skill and cognition in assessing mental activity from the ... · 2002 by the wenner-gren foundation...

Documents