Memory & Cognition 1980, Vol. 8 (5), 415423 Represeritation of inflected nouns in the internal lexicon
G. LUKATELA, B. GLIGORIJEVIC, and A. KOSTIC University ofBelgrade, Belgrade, Yugoslavia
and
M.T.TURVEY University ofConnecticut, Storrs, Connecticut 06268 and Haskins Laboratories, New Haven, Connecticut 06510
The lexical representation of Serbo-Croatian nouns was investigated in a lexical decision task. Because Serbo-Croatian nouns are declined, a noun may appear in one of several gram matical cases distinguished by the inflectional morpheme affixed to the base form. The gram matical cases occur with different frequencies, although some are visually and phonetically identical. When the frequencies of identical forms are compounded, the ordering of frequencies is not the same for masculine and feminine genders. These two genders are distinguished further by the fact that the base form for masculine nouns is an actual grammatical case, the nominative singular, whereas the base form for feminine nouns is an abstraction in that it cannot stand alone as an independent word. Exploiting these characteristics of the Serbo Croatian language, we contrasted three views of how a noun is represented: (1) the independent entries hypothesis, which assumes an independent representation for each grammatical case, reflecting its frequency of occurrence; (2) the derivational hypothesis, which assumes that only the base morpheme is stored, with the individual cases derived from separately stored inflec tional morphemes and rules for combination; and (3) the satellite-entries hypothesis, which assumes that all cases are individually represented, with the nominative singular functioning as the nucleus and the embodiment of the noun's frequency and around which the other cases cluster uniformly. The evidence strongly favors the satellite-entries hypothesis.
Inflection is the major grammatical device of Serbo see that most nouns were masculine and that the nomi Croatian, Yugoslavia's principal language. In general, the native singular was the most popular grammatical case. grammatical cases of nouns are formed by adding a The italicized numbers are normalized percentages and suffix to a root morpheme, where the suffix is of the can be read as follows, taking the masculine gender as an vowel, vowel-consonant, or vowel-consonant-vowel type. example. For any given masculine noun that occurs in Less frequently, inflection involves additional processes, the language with frequency f, the nominative singular such as vowel deletion and consonant palatalization. form of that noun occurs with a frequency of approxi The grammatical cases of Serbo-Croatian nouns mately .29f, the genitive singular form with a frequency produced by inflection are not equal in their frequency of approximately .19f, the dative singular form with a of occurrence. Table I summarizes the frequency frequency of approximately .02f, and so on. In short, analysis of D. Kostic (1965) on a corpus of approxi the normalized percentage for a given grammatical case mately 2 million Serbo-Croatian words appearing in of a given gender is the likelihood that, when a noun of the daily press and contemporary poetry. The non that gender appears, it appears in that particular case. italicized numbers are actual percentages. Thus, for all The question of interest to the present paper is how nouns in the corpus, 12.83% were masculine nouns in the inflected Serbo-Croatian nouns are represented in the nominative singular, 7.8% were feminine nouns in lexical memory. Following MacKay (1978) and Manelis the genitive singular, .13% were neuter nouns in the and Tharp (1977), we can distinguish two hypotheses instrumental plural, and so on. Reading the totals, we about the lexical representation of words with common morphological stems. According to the independent This research was supported in oart by NICHD Grant HD entries hypothesis, the individual grammatical forms of a 08495 to the University of Belgrade and in part by NICHD Serbo-Croatian noun would be represented in the lexicon Grant HD.{) 1994 to Haskins Laboratories. Reprint requests by independent representations, one internal representa can be sent either to G. Lukatela, Faculty of Electrical Engi neering, Bulevar Revolucije 73, Beograd, Yugoslavia, or to tion for each grammatical form. On the derivational M. T. Turvey, Haskins Laboratories, 270 Crown Street, New hypothesis, rather than all the forms of a given noun Haven, Connecticut 06510. being instantiated in the internal lexicon, there would be
Copyright 1980 Psychonomic Society, Inc. 415 0090-502X/80/050415-Q9$O 1.15/0 416 LUKATELA, GLIGORIJEVIC, KOSTIC, and TURVEY
Table I Case Frequencies in Percentages Singular Plural Masculine Feminine Neuter Total Masculine. Feminine Neuter Total Nominative 12.83 8.84 2.88 24.55 3.33 3.58 .69 7.60 28.89 22.56 20.44 7.50 9.14 4.30 Genitive 8.56 7.88 3.47 19.91 3.96 3.22 .61 7.79 19.27 20.11 24.63 8.92 8.22 4.33 Dative .87 .38 .31 1.56 .28 .16 .04 .47 1.96 .97 2.20 .63 .41 .28 Accusative 5.49 5.48 2.55 13.52 2.21 2.75 .73 5.69 12.36 13.99 18.10 4.98 7.02 5.18 Instrumental 1.90 1.94 .86 4.70 .60 .80 .13 1.46 4.28 4.95 6.10 1.35 1.86 .92 Locative 3.77 3.42 1.61 8.80 .61 .80 .21 1.62 8.48 8.73 11.43 1.37 2.04 1.48 Total 33.42 27.94 11.68 73.04 10.99 11.24 2.41 23.64 75.25 71.31 82.89 24.75 28.69 17.11 Note-This tableisadopted from D. Kostic (1965). Figures in italics represent the normalizedpercentages, asrelatedto the particular gender. Percen tages do not add to 100% because the rarely occurring vocative casehasbeenomitted. but one instantiation, probably of the noun's root successful (meaning that the letter string is a nonaffixed morpheme. There would also be in memory only a word), then a positive response can be initiated. How single instantiation of the set of inflectional morphemes. ever, if the search is unsuccessful, then the letter string Appropriate combinations of the root morpheme and is decomposed and the combination of root and affix inflections would be determined by separately stored is tested for its validity. Obviously, on the decomposition syntactic rules. second argument, lexical decision should be slower for There have been relatively few direct contrasts of affixed words. The Manelis and Tharp (1977) investi the two hypotheses for English lexical items, and the gation failed to find a difference between affixed and results have been largely equivocal. Manelis and Tharp nonaffixed words in either direction, a result that (1977) compared lexical decision ("Is this letter string favored the independent-entries hypothesis over either a word?") times for pairs of affixed words (words version of the decompositional hypothesis. consisting of two morphemes, a root morpheme and a However, the failure to find evidence for morpho suffix) with lexical decision times for pairs of nonaffixed logical decomposition with suffixed words contrasts words (words consisting of a single morpheme). Manelis with the provision of such evidence by Taft and Forster and Tharp (1977) predicted two possible outcomes from (1975) for prefixed words. These investigators reported the derivational or, as they termed it, decompositional that rejecting real roots (for example, SULTS as in hypothesis. For a given letter string, decomposition into INSULTS) as words took longer than rejecting false root and ending could be an obligatory first step, with roots (for example, NINGS as in INNINGS) as words. lexical search for the whole item a contingent later The interpretation given was that real stems. would be step; or lexical search for the whole item could be the found in the lexicon and a subsequent check would be initial obligatory step, with decomposition occurring needed to determine that these lexical entries do not later and dependent upon failure to find the whole item constitute words in the absence of an appropriate in memory. Consider the prediction that follows from prefix. the notion that decomposition occurs first. A word, A further demonstration of morphological decompo whether it be affixed or nonaffixed, is partitioned into sition is reported by MacKay (1978), although his exper root and ending. A test is then made to determine iment is distinguished from the experiment described the validity of the combination as an affixed word. If above in that it looks at the production of words rather the combination proves valid, a positive response is than at their perception. Subjects heard verbs (e.g., con initiated; if it proves invalid (meaning that the word is clude, decide) that they had to nominalize (conclusion, nonaffixed), a search of the lexicon is conducted for the decision) as rapidly as possible (MacKay, 1978). It was nondecomposed letter string. In brief, with everything shown that certain nominalizations took longer than else equal, the decomposition-first argument predicts others; precisely, the more complicated the derivational faster lexical decision for affixed words than for non process (the more steps intervening between verb form affixed words. The contrary prediction follows from the and noun form), the slower the nominalizations. decomposition-second argument. If the initial search of The source of the discrepancy between the experi the lexicon for the nondecomposed letter string is ments of MacKay (1978) and Manelis and Tharp (1977) INFLECTED NOUNS 417 could be relatively trivial, a matter of differences in the coding of word frequency in lexical memory can be methodology. On the other hand, the discrepancy might referred to as an interentry account, then its popular arise from a deep-seated difference between the kind of alternative can be referred to as an intraentry account, memory structure needed to recognize words and the for here the emphasis is not on an entry's position kind of memory structure needed to produce them. In relative to other entries but on the individual entry's the former case, the analogy that has come to be adopted sensitivity to linguistic stimulation. According to the is that of a dictionary: The internal representations of intraentry account, each lexical entry is a device for words are coded on orthographic and phonological accepting evidence about the presence of the word it principles and are accessed accordingly. But in the latter represents (see the logogen model of Morton, 1969, case, that of the requirements of production, the opposite 1970). In the case in which the word in question occurs analogy is not that of a dictionary but of a thesaurus very frequently, the evidence needed for detecting its (Labov, 1978): The internal representations of words presence is less, or, equivalently, the threshold of its are coded on semantic principles and should be accessed lexical entry is lower, than in the case in which the word accordingly; for in production, the problem is to locate in question occurs rarely. On this view, lexical search is a word that expresses a given meaning. parallel, and, in common with the interentry view, its Whatever the reason for the equivocality identified duration is inversely related to a word's frequency above, we should note that, with regard to the repre of occurrence. It is not so clear, however, how the sentation of inflected nouns, the independent-entries intraentry view accounts for decision time when no hypothesis and derivational hypothesis are not exclusive. lexical entry is to be found (see Coltheart, Davelaar, A third hypothesis can be entertained that combines Jonasson, & Besner, 1977). features of the first two. We refer to it, picturesquely, If there is an independent entry for each grammatical as the "satellite-entries" hypothesis. Here are its dis case of a Serbo-Croatian noun, then we might suppose tinguishing characteristics: (1) Each grammatical case of that lexical decision times for the grammatical cases of a noun has a separate entry in the lexicon; (2) the a given noun will vary in proportion to their frequencies nominative singular entry functions as the nucleus of of occurrence. In a previous experiment (Lukatela, the noun and it expresses the frequency of occurrence Mandie, Gligorijevic, A. Kostic, Savic, & Turvey, 1978), of the noun that it represents; (3) lexical entries of the we examined this prediction from the independent remaining grammatical cases cluster (relatively) uniformly entries hypothesis and found it wanting. Lexical deci about the nominative singular entry and are organized sion time was not related by a unique, constant multi among themselves and in relation to the nominative plier to the corresponding logarithms of the proportional singular by a (for now unspecified) principle other than frequencies of three grammatical cases. Rather, the frequency. In short, the lexical entries of the oblique decision time for one case, the nominative singular, was cases of a noun are satellites to the lexical entry of the significantly less than the decision time to either of the noun's nominative singular. other two cases (the instrumental singular and the The second characteristic of the satellite-entries dative singular), which did not differ from each other hypothesis reflects a common assumption of hypoth in terms of decision time, even though they differed in eses about lexical memory, namely, that entries in the frequency. We interpreted this observation as support lexicon express the frequency of the word they repre for either a derivational hypothesis or a hypothesis sent. We pursue that assumption in the remarks that consonant with the point of view that the nominative follow, because it figures significantly in the eventual singular is the nucleus entry about which the entries predictions we wish to make. for the other grammatical cases cluster uniformly. There are two fashionable interpretations of how a The experiment to be reported here contrasts the word's frequency of occurrence is coded in the internal satellite-entries hypothesis with the independent lexicon. The entries in lexical memory may be likened entries hypothesis on the one hand and with the deri to the files in a filing cabinet ordered according to vational hypothesis on the other. To anticipate, the out frequency of usage (Forster & Bednall, 1976; Rubenstein, come of the experiment favors the satellite-entries Lewis, & Rubenstein, 1971; Stanners & Forbach, 1973). interpretation of the lexical organization of inflected A word's frequency of occurrence is expressed in lexical nouns. memory by the location of its lexical entry. Thus, on The experiment takes advantage of two facts of the the filing-cabinet analogy, the entries for the most Serbo-Croatian language. First, the same letter pattern frequently occurring words are to be found at the front (and, therefore, phonetic pattern) can represent more of the cabinet (at the start of lexical search), whereas than one grammatical case. For example, the inanimate those entries for the least frequently occurring words are noun SERPA (nominative singular form), which means to be found at the back of the cabinet (toward the end "pot," is written as SERPE and pronounced identically of lexical search). On this view, lexical search is serial, in the genitive singular, nominative plural, and accusative and its duration is inversely related to the frequency cases. Where identities exist, the case frequencies can be of occurrence of the target word; when no lexical entry compounded. The case identities and their compound is to be found (i.e., when the letter string is a nonword), frequencies for nouns of the masculine and feminine the search is exhaustive. If the filing-cabinet account of genders are given in Table 2. , , 418 LUKATELA, GLIGORUEVIC, KOSTlC, AND TURVEY
Table 2 Identical Grammatical Cases and Their Compound Frequencies Masculine Nouns Percent Percent (Inanimate) Occurrence Feminine Nouns Occurrence Nominative singular, accusative singular 41.25 Nominative singular, genitive plural 30.78 Genitive singular, genitive plural 28.19 Genitive singular, nominative plural, accusative plural 36.27 Locative singular, dative singular 10.45 Locative singular, dative singular 9.70
The second fact to be exploited is that, whereas the entries hypothesis makes a considerably simpler pre nominative singular is the root morpheme in the declen diction: For both genders, the nominative singular will sion of masculine nouns, it is not the root morpheme in be responded to faster than the genitive singular. the declension of feminine nouns. For the latter, the The two hypotheses can be further contrasted with root morpheme is an abstraction in the loose sense that respect to their predictions on lexical decision times to the root morpheme never occurs asan actual grammatical the instrumental singular, which occurs with a propor case. In terms of distinctions sometimes used by lin tional frequency of approximately .04f in the masculine guists, the root morpheme of masculine nouns is full and approximately .OSfinthe feminine. The independent (it has semantic content) and free (it can stand alone as entries hypothesis would predict that decision times an independent word), whereas the root morpheme of to the very low-frequency instrumental singular of both feminine nouns is less obviously full, and it is certainly genders should be much longer than the decision times not free. Table 3 gives examples of the two genders. for the high-frequency nominative singular and the Let us return to the first fact identified above and put high-frequency genitive singular. The satellite-entries it to use as a means of prying apart the perspective of hypothesis, in contrast, predicts that lexical decision satellite entries from that of independent entries. The time for the instrumental singular should, in both compounded frequency of the nominative singular form genders, be very close, probably identical, to that for in the masculine gender proves to be greater than that of the genitive singular and significantly longer than that the genitive singular form in the masculine gender. For for the nominative singular. A summary of these con nouns of the feminine gender, this relation is reversed: trasting predictions of the two hypotheses is given in The nominative singular form occurs less frequently Table 4, in which the inequality symbols are in reference than the genitive singular. Thus, for a masculine noun to lexical decision time and the letters identify the of frequency of occurrence f, the respective propor nominative singular (ns), genitive singular (gs), and tional frequencies of the nominative singular and geni instrumental singular (is). tive singular letter patterns are approximately .41f The rationale for pooling the frequencies of visually and .28f. In contrast, for a feminine noun of frequency identical cases is that a reader's sensitivity (in lexical of occurrence f, the respective proportional frequencies decision) to a given grammatical form of a given noun is are approximately .31£ and .36f. The independent determined solely by the relative frequencies with which entries hypothesis would predict a shorter latency the reader has seen that grammatical form as a visual lexical decision for nominative singular masculine nouns object. A different perspective, however, and one that than for genitive singular masculine nouns. That same is more consonant with the satellite-entries hypothesis, is hypothesis, however, with respect to feminine nouns, that it is the visual form in a sentential context (i.e., as would predict either little difference in lexical decision a grammatical object rather than as a crass visual object) latency for the two grammatical cases or a difference in that is important, so that there are indeed separate which the decision time to the genitive singular form is lexical entries for individual cases that are visually the briefer of the two. In comparison, the satellite- identical but grammatically distinct. On this latter
Table l Declension of a Masculine Noun and of a Feminine Noun Masculine Feminine
Case Singular Plural Singular Plural Nominative DINAR DINARl ZENA ZENE Genitive DINARA DINARA ZENE ZENA Dative DINARU DINARlMA ZENI ZENAMA Accusative DINAR DINARE ZENU ZENE Vocative DINARE DINARl ZENO ZENE Instrumental DINAROM DINARIMA ZENOM ZENAMA Locative DINARU DINARlMA ZENI ZENAMA Note-Dinar =money; iena =woman. INFLECTED NOUNS 419
perspective, we should predict latency relations on the Table 6 basis of the uncompounded frequencies, as given in Predictions of Derivational and Satellite-Entries Hypotheses Table 1. The relevant predictions are shown in Table 5, Masculine Feminine and, as comparison of Tables 4 and 5 reveals, the pre Hypothesis Nouns Nouns dictions from compounded and uncompounded fre Decomposition Second ns < gs = is ns = gs = is quencies differ only slightly. Decomposition First ns>gs=is ns = gs = is Let us now take the second fact identified above, Modified Decomposition First ns = gs = is ns=gs=is namely, the differential status of the nominative singular Satellite Entries ns Three 3S-mm slides were constructed for each noun: one for the noun's nominative singular, one for the noun's genitive 800 singular, and one for the noun's instrumental singular. Accord ingly, there was a total of 162 slides in which the string ofRoman (see Lukatela, Savic, Ognjenovic, & Turvey, 1978) letters (Helvetia light, 12 point), arranged horizontally at the center I of the slide, spelled a word in Serbo-Croatian. I A set of 162 nonword slides was constructed by converting U 750 I each word from a different list of words meeting the same criteria as above into a nonword. This was done in the nomina i" tive singular and genitive singular cases by changing the first ... II I ~ letter and in the instrumental singular case by changing the last ;:: letter, so as to avoid idiosyncratic instrumental endings. II z 0 I Procedure ... 700 II V On each trial, the subject's task was to decide as rapidly as « possible whether the presented letter string was a word or a ... I nonword. Each slide was exposed for 1,500 msec in one channel '" I of a three-ehannel tachistoscope (Scientific Prototype Model GB) illuminated at 10.3 cd/rn". Both hands were used in responding I to the stimuli. Both thumbs were placed on a telegraph key 650 I close to the subject, and both forefingers were placed on another telegraph key 2 in. farther away. The closer button was depressed for a "no" response (the string of letters was not a word), and I the farther button was depressed for a "yes" response (the NOMINATIVE GENITIVE INSTRUMENTALI string of letters was a word). Figure 1. Reaction time to three grammatical cases of nouns Latency was measured from stimulus onset. The total session of the masculine gender (striped bars) and nouns of the feminine lasted 30 min, with a short pause after every 18 slides. gender. Design syllables. Word length is known to contribute signifi Each subject saw a total of 108 slides, of which 54 were cantly to response latencies (Whaley, 1978). words and S4 were nonwords, but no subject saw any given letter string or any given noun more than once in the course of The design of the present experiment was so chosen the experiment. This was achieved in the following marmer. The as to insure that no subject saw the same noun twice. S4 feminine and masculine nouns were divided into three groups It is a design, however, that raises certain difficulties (A, B, C) of 18 nouns each. The 60 subjects were divided into for situations in which one is concerned with keeping three groups (I, 2, 3) of 20 subjects each. Subjects in Group 1 the analysis true to the strictures advocated by Clark saw the nominative singular cases of Category A nouns, the genitive singular of Category B nouns, and the instrumental (1973), that is, of treating both subjects and letter singular of Category C nouns. Subjects in Group 2 saw the strings as "random effects" and computing reliability nominative singular case of Category B nouns, the genitive singu of results over both of these sampling domains. To lar of Category C nouns, and the instrumental singular of Cate circumvent these difficulties, we used a variation of a gory A nouns. For subjects in Group 3, the categories were C, A, B, respectively, for nominative, genitive, and instrumental. procedure that we have reported previously (see Lukatela, A similar partitioning into categories and mapping onto subject Savic, Gligorijevic, Ognjenovic, & Turvey, 1978). groups was done for the nonwords. A comparison within a gender between any two of the three grammatical cases is composed of two sub RESULTS comparisons: one in which the nouns are the same but the subjects are different (comparing decision times for Figure 1 gives a histogram plot of the mean reaction A words, B words, and C words), and one in which the times for the three grammatical cases of the masculine subjects are the same but the nouns are different (com and feminine nouns. Reaction times smaller than paring decision times for Group 1, Group 2, and 300 msec and greater than 1,500 msec were excluded Group 3). The two quasi-F ratios for these subcompari from the calculations of the means, as were erroneous sons are viewed as random variables, the probabilities responses, which occurred in the present experiment of which have a chi-square distribution with 2 by 2 degrees at a rate of less than 2.5%. Only the latencies to words of freedom. These new random variables are computed are considered in the analysis below. as Ii = -2 In (Pi) for any subcomparison Ii for which Inspection of Figure 1 suggests a difference in the the F' is at the probability level Pi. The obtained sum of rank order of grammatical caselatencies between genders. the new variables is then assessed for significance against At the same time, however, the figure does not suggest the chi-square value for the corresponding degrees of a pattern of results consonant with the predictions of freedom. In short, this analysis assesses the likelihood the alternatives to the satellite-entries hypothesis. A that a set of two quasi-F ratios with probabilities of difference between the genders might hold for the PI and P2 could have come about by chance. absolute latencies. The apparently slower overall response For the masculine nouns, the nominative singular to the masculine nouns might be due to their generally differed from both the genitive singular [X2(4) =28.65, greater length in both number of letters and number of p< .001] and the instrumental singular [X2(4) =19.44, INFLECTED NOUNS 421 p < .00 I], which did not differ between themselves it consists of a root form and an inflectional ending, but [X2(4) = 5.51, p > .05]. The same pattern held for the the nominative singular of masculine nouns is unlike feminine nouns [nominative singular vs. genitive singu other cases in that it is the root form and contains no lar, X2(4) = 29.46, p < .001; nominative singular vs. inflectional ending. Two versions of the derivational instrumental singular, X2(4) = 35.45, P < .001; genitive hypothesis (see Table 6) predict differences between singular vs. instrumental singular, X2 (4) =1.58, p > .05] . masculine and feminine nouns in the pattern of deci sion latencies among the grammatical cases. The experi DISCUSSION ment revealed, however, that the pattern for the two genders is the same, not different. A third version of The purpose of the present experiment was to assess the derivational hypothesis does predict identical patterns three interpretations of how the inflected nouns of for masculine and feminine nouns, but the predicted the Serbo-Croatian language are represented in the pattern is one in which there are no latency differences internal lexicon. On one interpretation, the independent among grammatical cases. We are reminded that for entries hypothesis, it is assumed that each grammatical both genders the experimental outcome was a latency case is stored in the lexicon as a separate and relatively difference that favored the nominative singular over independent entry. Insofar as an entry in the internal the other two cases. Thus the third version of the lexicon is believed to embody, either through its rela derivational hypothesis does not hold either. tion to the other entries or through its sensitivity to Before we draw any general conclusions from the linguistic stimulation, the frequency of occurrence of present data, it behooves us to consider an aspect of the the word that it represents, it should be argued that the design that might give reason for caution. The basis for grammatical cases of any given noun must relate among the fifth restriction on the choice of words described themselves in terms of their frequencies of occurrence. above, that the masculine nouns be inanimate, was This prediction of the independent-units hypothesis that in the declension of nouns of the masculine gender was examined through an investigation oflexical decision the grammatical cases that are visuallyjphonologically to three grammatical cases: the nominative singular, the identical are not the same for nouns denoting animate genitive singular, and the instrumental singular. The and inanimate objects. For example, the genitive singular relation between the first two cases differs as a function is identical in form to the genitive plural in the case of of noun gender: For masculine nouns the nominative inanimate nouns and identical in form to the genitive singular is of greater compounded frequency, whereas plural and accusative singular in the case of animate for feminine nouns the genitive singular is (on com nouns. For the compounding of frequencies, it seemed pounding identical grammatical cases) the more fre prudent to stay with just one kind of masculine noun, quently occurring form. In both genders the instru although either kind would have been adequate for the mental singular occurs far less frequently than the purposes of the experiment. However, in retrospect, other two cases. The pattern of lexical decision latencies our choice to consider only one of the two kinds of to be expected from the independent-units hypothesis masculine noun may have introduced an unnecessary was not realized; rather than there being one pattern complication. A native speaker of English unfamiliar for the masculine nouns and another for the feminine with Serbo-Croatian might intuit that the contribution nouns, there was a single pattern, the same for both of the animate and inanimate nouns to the relative genders. Importantly, lexical decision time was briefest frequencies of masculine grammatical cases given in for the nominative singular of both genders, and there Table I is not the same (for example, one kind of mascu was no latency difference between the genitive singular line noun might contribute more to the frequency of and instrumental singular of both genders. one case than to another), and, therefore, to select one The obtained results are consistent, therefore, not of the two kinds of masculine noun is to make void the with an independent-units hypothesis as we have inter use of the tabulated frequencies. preted it, but with a hypothesis that assumes that not In English, possession is marked by's. If this form is all grammatical cases are qualitatively alike in lexical taken as sole representative of the genitive case, then status and that the grammatical cases are not ordered given that the use of's tends to favor animate over among themselves according to frequency of occurrence. inanimate nouns, one might suppose that the genitive One grammatical case, the nominative singular, appears case is the hallmark of animate nouns. However, English to playa pivotal role due in part, perhaps, to its primacy combines inanimate nouns with the preposition "of' in acquisition (Carroll & White, 1973a, 1973b). The to produce effectively a partitive genitive: " ... of the latter fact is important in another way, too: It argues car," " ... of the paper" (see Jaspersen, 1962). It is against a derivational hypothesis in which lexical deci unlikely that these two kinds of genitive differ markedly sion involves successive stages of decomposing into the in their frequencies of occurrence. In Serbo-Croatian, root and inflectional morphemes and testing the com the genitive case, unlike its counterpart in English, is bination for its legality. Morphologically, the nominative a very complex case, assuming 13 different grammatical singular of feminine nouns is like all other cases, in that functions. Of these functions, one is exclusively related ,, 422 LUKATELA, GLIGORIJEVIC, KOSTIC, AND TURVEY to inanimate nouns (Stefanovic, 1974). As with English, the threshold of the nonnominative singular's logogen it seems unlikely that the frequency of the genitive incremented by a constant. It is, perhaps, in some such case in Serbo-Croatian would be significantly less for sense as this, in the way in which the thresholds of the inanimate nouns than for animate nouns. lexical entries for oblique grammatical cases are tied by a Similar comments need to be made in reference to constant to the threshold of the lexical entry for the the instrumental case, for here one might suppose that nominative singular, that we can begin to interpret the inanimate nouns take th! instrumental form more often intuitive notion of a satellite organization for the than animate nouns. In Serbo-Croatian, there are three inflected nouns of Serbo-Croatian. In view of the out categories of instrumental: instrumental case without come of the present experiment, we conclude that the preposition (eight kinds), instrumental case with the hypothesis of a nucleus logogen representing the nomi preposition "with" (three kinds), and instrumental case native singular and about which the logogens of the with spatial preposition (above, under, in front of, oblique cases cluster uniformly is a better candidate for between/among). Of these three types, only two are understanding the lexical organization of inflected nouns exclusively related to inanimate nouns (Ivic, Note 1). than either the hypothesis that the cases are represented Of course, the point we are trying to establish is that independently of each other or the hypothesis that they the case frequencies for masculine nouns as reported in are derived by rule. Table 1, on the basis of which we formed our predic Recently (and subsequent to the design and imple tions concerning the respective hypotheses of lexical mentation of the present experiment), a description of organization, are equally applicable to masculine nouns lexical organization has been proposed (Taft, 1979b) of both the inanimate and animate kinds. Nevertheless, that accommodates the features of both the independent in the absence of case-frequency norms for individual entries and the decomposition hypotheses. The lexicon words (which are not currently available), there is still is said to consist of a master file and a number of periph some room for doubt, although we believe it to be eral files: orthographic, phonological, and semantic small, that the foregoing contention holds. A small (Forster, 1976). In the master file, the surface form of empirical point in our favor is that the mean decision each word is separately and completely represented. In times of 39 subjects for 10 animate and 10 inanimate the peripheral files, on the other hand (of which the masculine nouns drawn from the stimuli of the previous orthographic is the one of special significance to visual experiment (Lukatela, Mandie, Gligorijevic, A. Kostic, word recognition), it is base forms that are represented Savic, & Turvey, 1978) were virtually identical for both rather than surface forms. Peripheral files store informa nominative singular and instrumental singular cases: tion that is sufficient for selectively and successfully 594 msec and 680 msec, respectively, for the 10 inani accessing the master file, where all information is to be mate nouns and 591 msec and 674msec, respectively, found. It is argued that in the orthographic file the first for the 10 animate nouns. If animate and inanimate syllable of a word, defined orthographically and mor masculine nouns differ markedly in the frequency with phologically, identifies the base form (Taft, 1979a) and which they occur in the instrumental case and if decision that the frequency of a given base form is defined by the latency reflected that frequency distinction, then the summed frequencies of the individual words of which it lexical decision times should have differed. is the first syllable (Taft, 1979b). Importantly, in both We would argue, therefore, that, taken collectively, kinds of file, master and peripheral, the frequency of an the present experiment and the previous one (Lukatela, entry is a significant determinant of access time. Mandie, Gligorijevic, A. Kostic, Savic, & Turvey, 1978) Consider the lexical representation of an inflected support the assumption that the oblique nonnominative Serbo-Croatian noun from the perspective of the master singular cases do not differ in relative accessibility due flle/peripheral file notion. There would be for a given to their differences in frequency of occurrence, but noun a single entry in the orthographic fIle-say, the rather that they are equally accessible. To date, we have first syllable-with a frequency determined by the noun's found little evidence for a difference in lexical decision occurrence in the language, and 14 entries in the master latencies among the genitive singular, locative singular, file (I entry for each grammatical case), with their and instrumental singular cases (and, therefore, in individual frequencies determined by the frequencies addition, among their visually identical mates; see of occurrence of the individual cases that they represent. Table 2). Given nouns such as ZENA and DINAR, the peripheral Suppose that, after Morton's (1969, 1970) logogen file would contain ZEN and DIN, respectively, whereas model, we assume that the lexical representation of the the master file would contain, for each of the two nominative singular has a threshold inversely propor nouns, the full form of each grammatical case. Lexical tional to the frequency with which the noun (indifferent decision occurs via these steps: First, the noun is decom to its particular grammatical case) occurs in the language. posed into the first syllable and afflxes. Second, a search Then, given the preceding observation, we should of the peripheral file is conducted for a length of time suppose that there is a common threshold level for the determined by the frequency of the base form. Third, logogens of the oblique cases that is at a value equal to the master file is accessed (through the address given by INFLECTED NOUNS 423 the base-form entry in the peripheral file), and the COLTHEART, M., DAVELAAR, E., JONASSON, T., & BESNER, D. legality of the base-form/affix(es) combination is ascer Access to the internal lexicon. In S. Dornic (Ed.), Attention and tained at a speed determined by the frequency of the performance VI. London: Academic Press, 1977. FORSTER, K. I. Accessing the mental lexicon. In E. C. T. Walker combination (i.e., by the frequency of the individual & R. J. Wales (Eds.), New approaches to language mechanisms. grammatical case). We see, in short, that although the Amsterdam: North-Holland, 1976. master file/peripheral file notion ascribes to the decom FORSTER, K. I., & BEDNALL, E. S. Terminating and exhaustive position hypothesis, it predicts the same outcome as the search in lexical access. Memory & Cognition, 1976, 4, 53-61. FORSTER, K. I. Levels of processing and the structure of the independent-entries hypothesis, namely, that decision language processor. In W. E. Cooper & E. C. T. Walker times are a function of the relative frequencies of the (Eds.), Sentence processing. Hillsdale, N.J: Erlbaum, 1979. individual grammatical cases. GLANZER, M., & EHRENREICH, S. L. Structure and search of the Our conclusion concerning the organization of internal lexicon. Journal of Verbal Learning and Verbal Behav ior, 1979,18,381-398. inflected Serbo-Croatian nouns, based as it is on the J ASPERSEN, O. Essentials of English grammar. London: Allen & indifference of decision latency to grammatical case Unwin, 1962. frequency, does not concur with the master file/peripheral KOSTIC, DJ. The structure of usage value of grammatical forms file notion, at least not with the current form of the in Serbo-Croatian. Report of the Institute of Experimental notion, for there are hints that distinct files is a needed Phonetics and Speech Pathology, Belgrade, 1965. LABOV, W. Gaining access to the dictionary. In J. F. Kavanagh conception for certain aspects of lexical access (e.g., & W. Strange (Eds.), Speech and language in the laboratory, Forster, 1979; Glanzer & Ehrenreich, 1979). Therefore, school and clinic. Cambridge, Mass: M.LT. Press, 1978. we would expect the general idea to receive further LUKATELA. G., MANDIC, Z., GLIGORIJEVIC, B., KOSTlc, A., attention and to undergo modification. One major SAVIC, M., & TURVEY, M. T. Lexical decision for inflected nouns. Language and Speech, 1978,21, 166-173. reason for the lack of concurrence may rest with the LUKATELA, G., SAVIC, M., GLIGORIJEVIC, B., OGNJENOVIC. P., issue of whether lexical organization is uniform or & TURVEY, M. T. Bialphabeticallexical decision. Language and pluralistic. Chomsky (1970) and others (e.g., Stanners, Speech, 1978,21,142-165. Neisser, Hernon, & Hall, 1979) have expressed a plural LUKATELA, G., SAVIC, M., OGNJENOVIC, P., & TURVEY, M. T. istic view arguing, for example, that the lexicon's organi On the relation between processing the Roman and Cyrillic alphabets: A preliminary analysis with bi-alphabetical readers. zational formats for the inflectional forms of English Language and Speech, 1978,21,113-141. verbs and for the nominal derivations of English verbs LUKIC, V. {Active written vocabulary ofpupils at the elementary need not be identical. And Bradley (1978) has given school age] (in Serbo-Croatian). Belgrade: Zavod za Izdavanje good empirical reasons for holding distinct the lexical Udzebenika SR Srbije, 1970. organizations of the closed set of words (often termed MACKAY, D. G. Derivational rules and the internal lexicon. Journal of Verbal Learning and Verbal Behavior, 1978, 17, function words) from the open set of words. Thus, the 61-71. fact that the affixed English nouns and verbs studied by MANELIS, L., & THARP, D. The processing of affixed words. Taft (1979b) and the inflected Serbo-Croatian nouns Memory & Cognition, 1977,5,690-695. studied by us submit to different explanatory accounts MORTON, J. Interaction of information in word recognition. Psy chological Review, 1969, 76, 165-178. of lexical organization may point less to an opposition MORTON, J. A functional model for memory. In D. A. Norman of data than to a differentiation of lexical organization (Ed.), Models of human memory. New York: Academic Press, according to differences in linguistic forms and functions. 1970. RUBENSTEIN, H. R., LEWIS, S. S., & RUBENSTEIN, M. A. Evi REFERENCE NOTE dence for phonemic recoding in visual word recognition. Jour nal of Verbal Learning and Verbal Behavior, 1971, 10,645-657. I. Ivic, M. {Case system in Serbo-Croatian language] (in Serbo STANNERS, R. F., & FORBACH, G. B. Analysis of letter strings in Croatian). Internal publication of the University of Novi Sad, word recognition. Journal of Experimental Psychology, 1973, Novi Sad, Yugoslavia, 1972. 98,31-35. STANNERS, R. F., NEISSER, J. J., HERNON, W. P., & HALL, R. Memory representation for morphologically related words. REFERENCES Journal of Verbal Learning and Verbal Behavior, 1979, 18, 399-412. BRADLEY, D. C. Computational distinctions of vocabulary type. STEFANOVIC, M. {Contemporary Serbo-Croatian language] (in Unpublished doctoral dissertation, Massachusetts Institute of Serbo-Croatian). Belgrade: Naucha Knjiga, 1974. • Technology, 1978. TAFT, M. Lexical access via an orthographic code: The basic CARROLL, J. B., & WHITE, M. N. Age-of-acquisition norms for orthographic syllable structure (BOSS). Journal of Verbal 200 picturable nouns. Journal of Verbal Learning and Verbal Learning and Verbal Behavior, 1979,18,21-39. (a) Behavior, 1973, 12, 573-576. (a) TAFT, M. Recognition of affixed words and the word frequency CARROLL, J. B., & WHITE, N. M. Word frequency and age of effect. Memory & Cognition, 1979, 7, 263-272. (b) acquisition as determinants of picture-naming latency. Quarterly TAFT, M., & FORSTER, K. Lexical storage and retrieval of pre Journal ofExperimental Psychology, 1973, 25, 85-95. (b) fixed words. Journal of Verbal Learning and Verbal Behavior, CHOMSKY, N. Remarks on nominalization. In R. A. Jacobs & 1975,14,638-647. P. S. Rosenbaum (Eds.), Readings in English transformational WHALEY, C. P. Word-nonword classification time. Journal of grammar. Waltham, Mass: Ginn, 1970. Verbal Learning and Verbal Behavior, 1978, 17, 143-154. CLARK, H. H. The language-as-fixed-effect fallacy: A critique of language statistic in psychological research. Journal of Verbal (Received for publication July 5, 1979; Learning and Verbal Behavior, 1973, 12,335-359. revision accepted March 12,1980.)