[Artigo] the Social and Structural Dimensions of Syntatic Change - Naro

Linguistic Society of America

The Social and Structural Dimensions of a Syntactic ChangeAuthor(s): Anthony J. NaroSource: Language, Vol. 57, No. 1 (Mar., 1981), pp. 63-98Published by: Linguistic Society of AmericaStable URL: http://www.jstor.org/stable/414287 .

Accessed: 02/06/2014 14:04

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

Linguistic Society of America is collaborating with JSTOR to digitize, preserve and extend access to Language.

http://www.jstor.org

This content downloaded from 200.129.202.130 on Mon, 2 Jun 2014 14:04:14 PMAll use subject to JSTOR Terms and Conditions

http://www.jstor.org/action/showPublisher?publisherCode=lsa

http://www.jstor.org/stable/414287?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp


THE SOCIAL AND STRUCTURAL DIMENSIONS OF A SYNTACTIC CHANGE

ANTHONY J. NARO

Pontificia Universidade Catolica and Universidade Federal, Rio de Janeiro

This paper proposes a model of syntactic change based on a quantitative study of the rule of subject/verb agreement in spoken Brazilian Portuguese. Among speakers of the lower socio-economic levels, this rule is currently undergoing a process of elimination from the grammar. The model postulates that syntactic change starts at a point where surface differentiation between the old and new systems is zero (or nearly so); later it spreads throughout the language in inverse proportion to the degree of saliency of the surface differences between these systems in each particular environment. Thus 'natural' clusters of linguistic features arise only gradually.

1. INTRODUCTION. Since syntactic change produces a finite number of dis- crete variants, it cannot be a process of gradual replacement of minutely dif- fering forms, operating unobserved over time.' In general, there is no surface continuum of realizations to be dealt with; rather, a certain (usually small) number of non-overlapping distinguishable variants can be identified and cat- egorized. This circumstance, which makes syntactic change an ideal site for observation of linguistic change-in-progress, has not been utilized in diachronic research. Even in recent times, historical syntax has been heavily inclined to a change-over-the-centuries methodology which makes it impossible to study the social and structural dimensions that shape the initiation and diffusion of a change; this has led to a distorted view of the process, in which only the endpoints are seen. Furthermore, since the usual endpoints are relatively sta-

'The central notion of syntactic diffusion along the axis of saliency, upon which the structural portion of this paper is based, was sketched in Naro & Lemle 1976. In that work, only three speakers, with a total of less than ten hours of recorded speech, were studied. Later, much more extensive tests, involving twenty speakers and 140 hours of speech, were made as part of the 'Competencias Basicas do Portugues' research project (sponsored jointly by the Fundacao Mo- vimento Brasileiro de Alfabetizafco and the Ford Foundation office in Brazil, administered by Miriam Lemle of the Museu Nacional, Universidade Federal do Rio de Janeiro). The results of these tests appear in the final report of the project (Lemle & Naro 1977:17-50). In the present paper, I use the same data base as in earlier versions, with only minor corrections and additions; but I report somewhat different results because of the inclusion here of social variables and some changes in the morphological categorization. I am grateful to Lemle for allowing me to use the data gathered during her project, as well as for many insightful discussions during the course of the work; without her keen intuitions into the workings of Portuguese, this research would not have been possible. I am likewise grateful to Gregory Guy, to whom I owe my initiation into the use of statistical techniques in linguistics, as well as many hours of fruitful collaboration; and to Shana Poplack and David Sankoff for comments and suggestions that have profoundly affected the final form of this paper. Michael Stanton of the Departamento de Informatica of PUC/RJ has patiently helped me learn how to communicate with a computer in a manner to which it will respond more or less appropriately, even if not always willingly and quickly. All calculations reported here were carried out on the IBM/370-165 installed at the Rio Datacentro, to which PUC/ RJ generously provided me free and unlimited access. Differences between this and previous versions reflect what I consider to be improvements or corrections.

63



LANGUAGE, VOLUME 57, NUMBER 1 (1981)

bilized standard languages, a deceptive appearance of categorical regularity is produced.

The goal of this article is to present a detailed quantitative study of one particular on-going syntactic change-the loss of 3rd person subject/verb concord in modern Brazilian Portuguese-and to suggest (on the basis of this and other evidence) a general hypothesis about syntactic change.

2. SUBJECT/VERB CONCORD. In standard Portuguese, a finite verb must agree with its subject, whether this latter element be present or deleted, preposed or postposed.2 Some prescriptive manuals do countenance a few exceptions to categorical agreement, such as optional singular or plural with subjects of the type um grupo de manifestantes 'a group of demonstrators'; but in the standard written language and in the normal speech of the educated classes, the rule of subject/verb concord is nearly categorical.

Within the noun phrase, number agreement also operates obligatorily in the standard language. As we shall see in more detail in ?9, below, this is not true of the spoken language-where, usually, only one element of the NP shows an explicit plural inflection. But in contrast to some varieties of Spanish, the presence of at least one explicit plural mark in the NP is nearly categorical, even in the most relaxed styles of the lowest socio-economic levels. This circumstance facilitates the identification of plural subjects. Generic NP's, which are more common in Brazilian Portuguese than in English, are gram- matically singular even in the standard language; e.g. Feijao preto ta em falta 'Black bean(s) is out of stock.' For this reason, they will not be considered here.

In the verb, pluralization is indicated by means of the person-number suffix- which, in the 3rd person, takes the basic forms -u or zero for the singular, and -un or -n for the plural (Camara 1972:126-7). On the surface, this suffix always occurs fused with the tense-mood suffix (and sometimes even with the theme vowel or root). Thus the actually occurring inflection can be complex; and it may show additional changes in form, resulting from the fact that the tense- mood suffix may also be dependent upon number (Camara 1972:125). To give a more complete view of the facts, I have listed in Table 1 the major possible surface realizations of the complete inflections for the 3rd person, along with examples and glosses. Both the standard orthographic representations and the phonetic forms of the inflections are indicated; the latter are further subdivided into popular and standard variants. The popular variants tend to occur more frequently in the speech of the lower socio-economic levels, while the standard variants are found more frequently in the speech of the higher socio-economic levels, on radio and television, etc.

A few irregular verbs for which singular and plural forms are homophonous (or nearly so) in the present have been excluded from the data base in that tense (e.g. ter 'he has', tem 'they have', where the accent is merely ortho-

2 In fact, even an infinitive must agree with its subject under certain circumstances. However, since inflected infinitives constitute an area of debate in both prescriptive and descriptive grammar, and occur very infrequently in spontaneous speech, they have been left out of consideration.

64



DIMENSIONS OF A SYNTACTIC CHANGE

CLASS

I. PRESENT 1. 1st conjugation

(regular) 2. 2nd & 3rd conj.

(regular) 3. 2nd & 3rd conj.

(-r or -z) 4. Monosyllables

II. PRETERIT IMPERFECT

All verbs

III. FUTURE All verbs

IV. PRETERIT PERFECT 1. 1st conjugation

(regular) 2. 2nd conjugation

(regular) 3. 3rd conjugation

(regular) 4. stress-shifting

(irregular) 5. radical changing

(irregular)

ORTHOGRAPHIC FORM

sg.: falA 'speaks' pl.: falAM 'speak' sg.: partE 'leaves' pl.: partEM 'leave' sg.: faz 'does' pl.: fazEM 'do' sg.: dA 'gives' p.: dAo 'give'

sg.: falavA 'spoke' pl.: falavAM 'spoke'

sg.: falarA 'will speak' pl.: falarAo 'will speak'

sg.: falou 'spoke' pl.: falARAM 'spoke' sg.: aprendEu 'learned' pl.: aprendERAM 'learned' sg.: partiu 'left' pi.: partIRAM 'left' sg.: trouxE 'brought' pl.: trOUXERAM 'brought' sg.: fez 'did' pi.: fiZERAM 'did'

TABLE I

POPULAR STANDARD

['a] ['fi, 'u, ,] ['i] [ I]

0 [-i, :i] [a] [aw]

['a]

[li] ['ey] 0

[a] [aw]

['a] ['a] ['u, 'u, 'a] [Iaw]

[-a] [-aw]

[-6] [-rfi, -aru] [-ew] [-eru, -dru] [-iw] [-frfi, -fru] [li] [-rfi, -dru] 0 [-erufi, -eru]

[-a] [-aw]

[-6, -6w] [-araw] [-drAW] [-dw] [-draw] [-fw] [-fraw] [-ri]

0 [-eraw]

graphic). Also excluded from consideration are several constructions in which it is not clear that a surface subject is present. For example, according to the prescriptive grammars used in Brazilian schools to teach the most prestigious forms of the standard language, the copula in time expressions should agree with the NP that denotes the hour; thus Era (sg.) uma hora quando Giselle chegou 'It was one o'clock when Giselle arrived'; Eram (pl.) duas horas quando Ant6nio chegou 'It was two o'clock when Anthony arrived.' However, in this construction, the time might be considered a predicate element, and the subject as empty. Similarly in clefts: Eram (pl.) tres mofas que vimos 'It was three girls that we saw.' Since these types of sentence show great variation even in the standard, because of varying syntactic construals, it would be inappropriate to include them in a study of agreement as such.

Finally, this study is limited to verbal agreement with true plural subjects, disregarding impersonal subjects of the type in Dizem que nao ter mais vaga 'They (people) say that there is no room left.' Earlier work (e.g. Naro & Lemle 1976) showed that these impersonal subjects impose a higher rate of plural marking than do true plurals. However, they are excluded here because of several peculiarities in behavior that are not directly relevant to the central theme. The data thus excluded amount to about 14.6% of the total corpus originally collected.

65




3. THE POPULATION UNDER STUDY. As mentioned at the outset, subject/verb concord is nearly categorical in the educated classes. The only circumstances under which large-scale variation occurs in this group is when the subject is postposed to the verb. Thus it was decided to limit the study to people of lower socio-economic levels, where variation is massive. Furthermore, since the principal sponsor of the research was MOBRAL (the Brazilian government's literacy agency), all but one of the seventeen people whose speech was to be studied were chosen from among students in that agency's classes. At the time of the interviews, they had completed about half of the six-month course. Some were repeats, having failed or dropped out of the previous cycle-but all were, for practical purposes, still essentially illiterate.3 In order to eliminate the geographic variable, all the speakers chosen were natives of the state of Rio de Janeiro, or had moved there as infants. Eight of them had spent their formative years in the surrounding areas outside the city proper, while nine were natives of the city itself. The speakers included eight females and nine males. They were concentrated in two disjoint age groups (roughly, 25 years or younger and 35 years or older) in order to provide a clear test of the distribution with respect to this variable; but there are only six speakers in the older group. The principal facts regarding each person are listed in Table 2.

NAME SEX AGE AREA PROFESSION

VAN F 15 S maid JOM M 16 C unemployed LID M 16 C errand boy SID M 16 C odd jobs HEN M 17 S helper (laundry) PAU M 19 S errand boy NAT M 20 C (unknown) SON F 22 C maid FAU F 23 S custodian (school) IZE F 23 C maid MAR F 26 C maid GCL M 37 C janitor UBI M 42 C janitor ILD F 43 S maid ELV F 45 S maid MAD F 45 S maid CAN M 54 S porter (club)

TABLE 2.

(M = male; F = female; C = city; S = state [environs])

Since syntactic data occur with a relatively low frequency in natural speech, it was decided to record about seven hours of speech for each person, in roughly hour-long interviews. However, NAT was interviewed for only one hour. The interview site, the interviewer, and topics under discussion were all varied as much as possible in an attempt to elicit the full range of each person's

3 GCL, the one non-MOBRAL student, is also illiterate.

66




potential behavior. Most of the interviews were carried out by university students, using a questionnaire designed primarily to elicit information on the speaker's background. Several topics related to popular culture were also suggested, but the interviewers had instructions to switch to any other topic which the interviewee seemed interested in discussing. In general, interviewer and interviewee were of the same sex. In some cases, one of the people in the sample interviewed another, and there were even attempts to obtain secret recordings. The total quantity of relevant data obtained varied greatly, however, ranging from 42 to 818 tokens4

4. MODELS AND METHODOLOGY. In a given sentence with plural subject, a variety of forces (sometimes called 'variable constraints') operate simultaneously to make the application of the agreement rule more or less probable, resulting in the variable appearance of a plural inflection on the corresponding verb. It is precisely the identification of such forces and the evaluation of their relative importance for rule application that constitute the central topic under investigation here.

For purposes of analysis, variable constraints may be classified into two main types: linguistic and social. The first is dependent upon relations that the item under study bears to others within the grammatical system; the second depends upon the relations of the speaker to the surrounding community. I shall postpone the study of social dimensions until ?6.

The basic determinant of the linguistic constraints can be deduced from comparison, in their spoken forms, of pairs of sentences like Eles come banana cor mel 'They eats (sg.) bananas with honey' and Eles e do Nordeste 'They is (sg.) from the Northeast.' The first, despite its lack of agreement, does not sound like an unusual utterance, while the second attracts attention much more strongly, and even sounds somewhat odd to many native speakers. Note, in this regard, that if the verb of the first example were pluralized, the resulting form comem 'eat' would not be very different phonetically from come 'eats': in the popular variants overwhelmingly preferred in the corpus under study in this article, the difference is merely nasalization of an unstressed final [i]. In the second sentence, however, the singular e 'is' is totally different from the plural sao 'are'. In other words, the second opposition is more SALIENT or noticeable than the first because the phonetic segments that realize the opposition are more different. On the basis of these rough observations, a multi- leveled scale of phonetic saliency of the morphological realization of the singular/plural opposition was set up, and was postulated as one of the structural constraint groups.

Along these same lines, lack of agreement is more salient when the plural subject occurs immediately preposed to the corresponding verb than when it is separated from the verb by intervening elements (adverbs, relative clauses

4 A 'relevant' token is defined as any occurrence of a verb with a plural subject, whether or not the verb actually has a plural inflection. Furthermore, it is not required that the plural subject be explicit in the sentence.

67




etc.) which physically and temporally separate the dependent plural verb from the determining plural subject. Even less salient is the case in which the subject follows the verb.

The effects of both the phonetic and the positional scales of saliency will be treated in greater detail below. I refer to these scales as 'constraint GROUPS' because they are set up in such a way that at'most one (in fact, exactly one) category of each group is applicable to a given verb.

4.1. UNIFORMITY OF EFFECT. As we have already seen, any variable phenomenon is under the influence of a number of simultaneous variable constraints. Even if the constraints relevant to the phenomenon under study were successfully identified, a seemingly unresolvable dilemma would remain: we wish to know the individual contribution of each constraint to the total situation, but we can never empirically observe only ONE constraint in action. In the real world, a certain number of constraints are always operating in any given environment. Thus, in his classic study of the deletion of the copula in Black English, Labov 1969 established that this rule was constrained by the following factor groups: (1) the nature of the subject (pronoun or full NP); (2) the nature of the predicate (NP, locative, predicate adjective, gerund, or gonna); and (3) the phonetic type of the preceding segment (sibilant, other consonant, or vowel). But the only available empirical data are of this type: With an NP subject ending in a vowel and with a gerund in the predicate, deletion occurs with frequency f. How, then, can we evaluate the contribution of the fact that, say, a gerund is present, if there is always some effect other than that of the gerund in any environment containing this element?

Note first that, merely by asking this sort of question, we are making the implicit assumption that it makes sense to talk about THE effect of a given element; i.e., we are assuming that whatever the effect of a following gerund may be, it is the same in all environments, regardless of whatever other elements may be present. This is, in general, not a self-evident truth. A simple example to the contrary can be given from NP agreement (Scherre 1978; discussed in greater detail in ?9, below). Here there is a strong tendency for the plural morpheme to be absent in all elements of the phrase following a non- agreeing element (i.e., for 'my first friends', as minhas primeira amiga is more common than as minha primeira amigas, where -s is the plural morpheme and as is the definite article). However, the effect of lack of -s is just the opposite of this if it occurs in the first element of the phrase (i.e., a minhas amiga is more common than a minha amigas, although both are rare). Thus it makes no sense to speak simply of 'the effect of a preceding uninflected element'. This problem becomes even more serious when social factors are introduced, since it is quite possible for the effect of a certain feature to vary among social groups (Kay 1978, Kay & McDaniel 1979).

When situations of this type are encountered in the course of research, two options are available: (1) to reformulate the variable constraint that violates the uniformity of effect assumption, or (2) to carry out separate analyses for each domain in which this assumption is satisfied.

68




Given uniformity of effect, at least within acceptable limits, we still need a model of the way the effects of each constraint combine to produce the total effect of the environment in which they are jointly present. The necessity for such models has, of course, arisen in other empirically-based sciences; the statistical solutions already found can, fortunately, be adapted for linguistics.

4.2. MATHEMATICAL MODELS. Basically, there are three approaches. The first ignores the whole problem by assuming that, if a sufficient number of data involve a given constraint c(i) in a wide variety of combinations with other constraints, the over-all result will be for the individual effects of the latter to cancel each other out. Under this assumption, it would suffice simply to note the frequency of rule application over all tokens in which c(i) is present (regardless of whatever other constraints might also be present), and then to take this raw frequency as a measure of c(i)'s effect on rule application. This sim- plistic approach does, in fact, work in many cases; but since its success depends more upon a felicitous distribution of the data than upon any factor that can be controlled, it would not be prudent to base any important conclusions on the results of this model alone. Furthermore, skewed distributions are usually unavoidable in natural linguistic data, since we have no way of controlling with precision the distribution of environments in the flow of speech.

The second approach postulates that the effects of each factor present in a given case cumulate additively, i.e. that the total effect of an environment is the SUM of the individual effects of the parts. Since we are here dealing with relative frequency of rule application, expressed as a percentage, this amounts to assuming that each variable constraint is associated with a certain quantity- also stated as a percentage, either positive or negative, that measures the effect of this particular constraint. According to this model, the experimentally determined percentage of rule application in any given environment should tend, within the limits of statistical fluctuation, to be the sum of these percentages. To provide a non-arbitrary standard against which to measure these effects, the grand mean, or over-all frequency of rule application, is added into the sum; the effects are then viewed as deviations from the grand mean resulting from the presence of the factors. In symbols, this model is:

(1) f(t) = f(O) + f(l) + ... + f(n) Here f(t) is the total frequency of application in a given context, f(O) is the grand mean, and f(l) ... f(n) are the effects of the n factors present in the context, stated as deviations from the grand mean.

A statistical technique known as multiple classification analysis (MCA) has been used to estimate the factorial deviations from the raw relative frequencies, by adjusting the raw deviation of each factor for the effects of the other factors in such a way as to satisfy formula I as closely as possible. The result is called an 'adjusted deviation'. MCA is usually used in conjunction with analysis of variance, the purpose of which is to determine whether the postulated factors have a statistically significant effect. Until recently, neither of these techniques was usable in areas of study where experiments could not be predesigned and controlled, because their mathematics required that the total number of tokens

69




in each environment be the same.5 Fortunately, improvements now allow this restriction to be relaxed under certain conditions. One advantage of MCA is that it provides an explicit measure of the relative importance of each GROUP of factors. In non-technical terms, this statistic, called a 'partial beta', indicates (but does not measure exactly), when squared, the proportion of the total variation attributable to the group to which it refers. Thus the betas vary between zero and one, and a higher value indicates a more important group.

In linguistics, the use of an additive model was first suggested by Labov 1969; but this suggestion was never implemented in any meaningful way, and Labov himself abandoned the additive model in his 1972 revision. The pre- mature demise of this model was apparently caused by a combination of the practical difficulties mentioned above (although the first version of the OSIRIS package, available in 1969, contained an appropriate computer program that did not require equal cell counts) with a misinterpretation of the underlying assumptions of the model. In the first place, Labov was forced to make a series of unrealistic assumptions (labeled the 'postulate of geometric ordering') in order to rule out the possibility of the sum in formula 1 becoming greater than 100% or less than 0%-or, in terms of relative frequencies (rather than percentages), greater than 1 or less than 0. However, there is no reason to constrain the model in this way: if, for a particular environment, the factorial effects estimated on the basis of all the data happen to add up to 110%, when the empirically determined rate is 95%, this should simply be counted as a failure to model the data correctly, in a way parallel to a prediction of 90% when the correct figure is 75%. In both cases, the proportion of variation accounted for will diminish and the error will increase. Unfortunately, later researchers (Kay & McDaniel, 160, 166-8) imposed a host of even more arbitrary and unnec- essary assumptions, since Labov's original 'postulate' turned out not to be sufficient for the purpose for which it was intended. All these unfortunate assumptions had the result of making the additive model quite unattractive.

One other aspect of the additive model should be mentioned. We have already noted above that the effect of each feature must be assumed to be constant, as a prerequisite to any discussion that makes sense. In the additive model, this has the consequence of requiring that each constraint add (or subtract) an equal amount to the frequency of rule application in all environments in which it is present. Consider an example from Labov's data (as reported by Kay & McDaniel, 165): if contraction of the copula in the environment of a preceding vowel occurs 90% of the time before a verb, but only 77% before a non-verb, the effect of the following non-verb is - 13%. If, in the environment of a preceding non-vowel, the frequency of contraction when a verb follows is 75%, the model thus requires that this frequency be 75% - 13% = 62% when d non- verb follows. To the extent that real data do not conform to such requirements, there is said to be INTERACTION among the constraints; i.e., at least one of the

I.e., if a speaker happened to utter n tokens of 1st conjugation verbs in the present tense with immediately preposed plural subjects, he/she would have to utter exactly the same number of tokens under identical circumstances in the preterit perfect, and so on.

70




constraints has an effect that varies according to the environment in which it is found. In Labov's data, the empirically determined frequency was 47%, rather than 62%; so apparently some interaction was involved.

If interaction effects are strong, the model in formula 1 will not be realistic, and the MCA based on it will be meaningless. Analysis of variance can be used to test for statistical significance of interaction; but statistical significance does not necessarily imply linguistic significance, since even minute interaction will be statistically significant with large amounts of data. The only practical way out of this difficulty is for interaction effects, if present, to be lumped in with the general error term. If the proportion of variation accounted for is nonetheless sufficiently high, the relegation of interaction to error is justified. This expedient has been used in the additive results presented below-which represent, as far as I know, the first successful use of the additive model in a linguistic study. This model will, however, always be used in conjunction with other models that have already proved their usefulness in previous linguistic investigations.

The third approach utilizes a multiplicative model for the combined effect of the component features of an environment. In this case, it is customary to speak of relative frequencies in the strict sense rather than percentages (i.e. without multiplying by 100), and to speak of the factorial effects as probabilities. This is because the first multiplicative models proposed in linguistics followed the usual law of independent joint effect of probabilities (Cedergren & Sankoff 1974:338-9):

(2) p(t) = p(O) x p(l) x *. x p(n)

Here the p(i) (for i from 0 to n) are in the interval between zero and one; p(t) represents the probability of rule application, corresponding to the empirically determined relative frequency up to statistical fluctuation, for an environment that consists of factors 1 to n; and p(O) is an 'input' probability that plays the same role as the grand mean in formula 1. The factorial probabilities are written as p(l), p(2) etc.

Note that this model has the unfortunate consequence that the combined effect of high probabilities (say, 0.8 and 0.9) is lower than their individual effects (0.8 x 0.9 = 0.72). Furthermore, this model also requires a series of unmo- tivated and unattractive assumptions, such as setting equal the probabilities associated with all factors that happen to occur in the environment in which the rule is most frequently applied. And, although it is true that 2 formally defines independence of application probabilities, we might just as well have postulated independence of non-application probabilities, replacing p(i) by (1 - p(i)). This introduces a further element of arbitrariness, since it gives us two equally well-motivated multiplicative models from which to choose. Fur- thermore, the basic reason for adoption of 2 disappears. For these reasons, the simple multiplicative model has proved inadequate and has been replaced by the so-called logistic model. Developed by statisticians for use in other areas, this model was introduced into linguistics by D. Sankoff 1975, and has been used with great success in a number of research projects. The basic formula

71




is:

p(t) (O) p(l) p(n) (1 - p(t)) (1 - p(O)) (1 - p(l)) (I - p(n))

The symbols here have the same interpretation as in 2, and continue to be called probabilities-although the appropriateness of this terminology is ques- tionable, since the logistic model is motivated merely by the fact that it works, rather than by any abstract considerations of probability theory. Representing the entire right-hand side of 3 by p(R) and solving for p(t), we find:

(4) p(t) = p(R) / (1 + p(R)) From this it is easy to see that, as p(R) becomes large, p(t) tends toward the number one; and as p(R) becomes small, p(t) tends toward zero. Since p(R), in turn, is a product of terms of the form

(5) p(i) I (1 - p(i)), p(R) will increase for each such term larger than 1, and decrease for terms less than 1. These terms themselves will be greater than 1 if p(i) is greater than 0.5, and less than 1 if p(i) is less than 0.5. If p(i) is exactly 0.5, the corresponding term becomes 1, and makes no contribution to the product. Thus, in the logistic model, the 'probabilities' can be classified as favoring, inhibiting, or indifferent-depending on whether they are greater than, less than, or equal to 0.5. Furthermore, according to this model, the combined effect of favoring factors (with values of 0.9 and 0.8, say, as in the example given above) is more highly favoring than that of each of the individual factors (the result is 0.97 in our example).

In the mathematical sense, the data and the model in 3 are not sufficient to determine uniquely the values of the p(i)'s; only a relationship among these probabilities is determined. In order to calculate a unique numerical value for each p(i), we must add one mathematical constraint per factor group. Although this constraint is, from the mathematical viewpoint, strictly arbitrary, it can be chosen in such a way as to facilitate interpretation of the resultant numbers. In any case, since all logistic results obey the same constraint, it will ordinarily not be necessary to consider its consequences. The constraint currently used is to require that the average of the terms in 5 for each factor group be zero. For groups containing only two factors, this has the consequence of requiring that the logistic probabilities sum to 1.

In order to estimate the values of the p(i)'s satisfying 3 (and the constraint just mentioned) on the basis of the available data, a statistical technique known as maximum likelihood estimation is employed. For any given model, the likelihood statistic itself is a measure of how likely it is that a particular set of data has been generated by the model. The estimation procedure maximizes this likelihood for the set of data under study, in accord with the model being used.

Within the limits of a particular model and data set, the significance of the results provided for any factor or factor group by the estimation procedure can be measured by eliminating the p(i)'s associated with such factors or groups, and verifying whether this has a significant adverse effect on the resultant

72




maximized likelihood. For example, let us return to Labov's study of copula deletion: we might suspect that, in the factor group referring to the nature of the predicate, the effect of a gerund is really not significantly different from that of gonna-or we might suspect that this whole group is irrelevant, i.e. that all its component factors really have the same effect, and that any apparent differences are statistically invalid. To test for statistical significance along these lines, we need only find the difference of the logs of the likelihoods for the two analyses, since twice this difference should approximate a chi-square distribution if no significant effect results from the additional factors of the more complex model. The number of degrees of freedom involved in the distribution is the number of factors eliminated from groups-or, if an entire group is eliminated, one less than the number of factors in the group. (For more details, see Rousseau & Sankoff 1978 and the sources cited there.)

Only highly sophisticated statistical techniques involving rigorous measures of the closeness of fit of the models to a wide range of empirical data could determine which (if any) of them is to be preferred in general. Fortunately, this problem does not exist for the phenomenon studied here-since it turns out that, with certain explainable exceptions, the three models give parallel results. Furthermore, since the models are intended only as approximations to joint effect, and are neither theoretically nor empirically justified in detail, we will be concerned more with relative strength of constraints than with exact numerical values.

4.3. CODIFICATION AND COMPUTATION. The tape-recorded interviews were transcribed by the staff of 'Competencias Basicas do Portugues', who located and marked verb tokens with plural subjects. Each such token was classified in accord with the structural constraints present in its environment (i.e. morphological category of the verb and position of the subject). For computational purposes, a single alpha-numeric symbol was assigned to each variable constraint (N for the morphological class offala/falam, A for immediately preposed subjects), and a tally was made of the number of times agreement applied in each environment. This resulted in a series of observations of the type 'Speaker VAN applied agreement 37 times out of 54 in the environment symbolized NA.' After a third alpha-numeric symbol was added to identify the speaker, these data were punched on cards in a standard format.

The cards were then submitted to SWAMINC4 (Naro 1978), an editing program that checks for errors of numerous sorts, removes data containing categorical (or 'knock-out') features, and provides a frequency count in accord with the first approach outlined above. Results based on the additive model were obtained by using the ANOVA subprogram of the SPSS (version 7) package (Nie et al. 1975:411-29); corresponding results according to the logistic multiplicative model were provided by a modified version of Sankoffs VAR- BRUL2 program that also calculates the log likelihood.

5. LINGUISTIC DIMENSIONS: RESULTS FOR MORPHOLOGICAL CATEGORY. AS stated in ?4, the basic observation that guided the ordering of the morphological categories was that increasing phonic salience of the singular/plural opposition

73




seemed to correlate with increasing chances of agreement. In order to reflect this central idea, a hierarchy of categories was set up so as to reflect increasing oppositional saliency. This morphological hierarchy was divided into two levels. The first level contains those pairs in which the phonetic segments that realize the opposition are UNSTRESSED in both members; the second level contains those pairs in which these segments are STRESSED in at least one member of the opposition. Given this fundamental distinction in terms of stress, it is clear that all oppositions on the first level are less salient than any on the second level.

The first level of the hierarchy was further divided into three classes, arranged in order of increasing material differentiation of the singular/plural opposition. In Class la, which consists of the present of the 2nd-3rd conjugations, this opposition is normally realized in our sample as [i]/[i]; i.e., the plural shows only nasalization, with no change in quality. Class lb, containing the present of the 1st conjugation, usually shows a more salient opposition of the type [a]/ [fi], where the nasalization in the plural is variable, but the differentiation in quality of the vowels serves to distinguish the forms. However, a few speakers have [a]/[a] for this opposition, and others show considerable variation among [a u] when the plural is utilized. In Class Ic, which contains the present of 2nd-3rd conjugation verbs ending in -r or -z, the entire unstressed segment is missing in the singular, resulting in a zero/[i] or zero/[i] opposition.

On the first level of the hierarchy, the opposition of Class la consists of the presence or absence of the suprasegmental feature of nasalization; Class lb shows an opposition marked by a greater number of distinctive segmental features; and Class Ic shows addition of an entire segment. Thus oppositional saliency increases through the three classes.

At the second level, we enter the range of the hierarchy in which the desinences are stressed. On this level, too, we distinguish several classes, and ar- range them in order of increasing material differentiation of the opposition. The least salient plural on this level shows nasalization of the stressed vowel, which does not change, and addition of an offglide. This occurs in Class 2a, containing the present of several common monosyllables (e.g. ddldao 'gives/give'),6 as well as the future of all verbs (e.g. falardlfalarao 'will speak [sg./pl.]') Class 2b, the next most salient, also shows no change in the stressed vowel, but an entire syllable, consisting of two segments, is added in the plural, contrasting with an offglide in the singular. All pairs in Class 2b are in the preterit perfect, but there are several distinct varieties. In the 2nd conjugation, the constant vowel is [e], and the opposition is [-ew/-eru] (comeu/comeram 'ate [sg./pl.]'), while the 3rd conjugation has [i] and [-iw/-iru] (partiulpartiram 'left [sg./pl.]') One unique case with constant vowel [o] and singular offglide [y] (foilforam 'was/were' or 'went [sg./pl.]') is also included in Class 2b.7 Class 2c, in addition to the syllabic contrast of 2b, shows a change in the stressed vowel from [o] to [a] in the opposition [-6/-ari] (falou/falaram 'spoke [sg./pl.]') On the second

6 Although slightly different, the opposition vailvao 'goes/go' is also included here.

7 Thus Class 2b combines the classes labeled J and Z in Lemle & Naro.

74




level of the hierarchy, then, the opposition of Class 2a consists of the presence or absence of nasalization (accompanied by an offglide); Class 2b shows an opposition marked by a greater number of segments, although the stressed segment remains constant; and in Class 2c even the stressed segment changes. Thus, oppositional saliency increases through the three classes of this level as in the first level.

As stated, Class 2c shows no shared segment between the stressed desinences that realize the singular/plural opposition, although the verb root is constant. The present of the verb ser 'to be' (el/so 'is/are') is unique: it is a monosyllable in which no clear separation of root from desinence is possible-and in which, as in the stressed desinences of Class 2c, singular and plural have no shared segment. To test the behavior of this unique case against that of Class 2c, it was assigned to a special Class 2d.

Finally, stress-shifting and radical-changing preterits (types 4 and 5 of Table 1) are included in Class 2e.8 In these verbs, the changing position of the accent introduces a high level of saliency; but the fact that there is no stressed desinence in the singular reduces the over-all effect.

A summary of the hierarchy of oppositional saliency, with the morphological classes arranged in accord with the levels of saliency discussed above, is as follows:9

(6) LEVEL 1 (unstressed): Class a. [-i/i]

b. [-a/iu] c. [-0/-]

LEVEL 2 (stressed): Class a. [-a/-aw]

b. [-ew/-eru,-fw/-fru, O-6y/-6ru] c. [-6/-aru] d. unique case: elsao e. ['0/-eru,-i/-eru]

8 Since these forms occur quite infrequently in real speech, it was not possible to maintain them as separate classes. The verb 'to come', which in the prescriptive standard has the infinitive vir and the preterit perfect veiolvieram 'came (sg./pl.), is currently undergoing a complex series of changes involving the introduction of nasalization throughout the paradigm, possibly in order to avoid confusion with ver 'to see' (see Camara [1965] 1975:89), which has similar or identical forms in its paradigm (e.g. vimos 'we saw' or 'we come'; veem 'they see' and vem 'they come', where the difference is merely orthographic). This process has resulted in the creation of a new infinitive vim [vi] for 'to come' (all other infinitives end in -r, but the most frequent spoken realization of this segment is zero), which happens to coincide with lsg. preterit perfect vim 'I came'. In the 3rd person preterit perfect, the new form is vinheram, replacing standard vieram 'they came'; but 3 sg. veio 'he came' is apparently remaining stable. Because of this situation of change in the plural, younger speakers of both groups overwhelmingly prefer to use the stable veio, tending to avoid both vieram and vinheram. This produces a decrease in the frequency of agreement for 'to come' in the preterit perfect, independently of the general rule under study here; hence data from this verb have been removed from Class 2e for the younger speakers, by means of data-exclusion commands of SWAMINC4 and SPSS.

9 The final vowels of Classes Ib, Ic, 2b, 2c, and 2e may be oral.

75




The results obtained for the morphological variable according to all three models are shown in Table 3. These results, as well as all others given in this section, were calculated for the group of 17 speakers as a whole; i.e., individuals were aggregated into a group (for justification, see ?7 below). The column labeled FREQUENCY reports the raw rate of agreement in accord with the first approach outlined above. For the logistic model, the factorial probabilities of agreement calculated by means of VARBRUL2 are listed under the heading PROB; the corresponding input parameter is 0.41. The results according to the additive model are cast in the form of deviations from the grand mean, or over- all frequency of agreement, resulting from the presence of the factors. These deviations, calculated by the MCA statistics option of the ANOVA subprogram of SPSS, are listed under the label ADJ DEV (adjusted deviation); the grand mean is 3002/6310, or 47.6%. In all instances, the ANOVA analysis itself showed that all main effects, as well as most interactions, were statistically significant at the .001 level; but interaction was included in the error term, as outlined above.

CLASS EXAMPLE FREQUENCY PROB ADJ DEV

la comelcomem 110/755 = 14.6% 0.07 -33.7% b falalfalam 763/2540= 30.0% 0.19 -18.2% c fazlfazem 99/273 = 36.3% 0.26 -11.3%

2a ddldao 604/927 = 65.2% 0.58 +17.9% b comeulcomeram 266/365 = 72.9% 0.69 +25.5% c faloulfalaram 524/672 = 78.0% 0.78 +33.3% d elsao 539/662 = 81.4% 0.79 +33.6% e disse/disseram 97/116 = 83.6% 0.80 +35.4%

(Beta = 0.74.) TABLE 3.

Inspection of the figures reported in this table shows that increasing rate operation of the agreement rule indeed correlates with increasing oppositional saliency. In all three models, the largest jump in rate of agreement occurs between levels 1 (stressed) and 2 (unstressed) of the oppositional hierarchy. This jump amounts to about 30% in terms of either raw or adjusted frequencies, and about 0.3 in terms of probabilities. Within Level 1, the least agreement is shown by Class la (nasalization, i.e. minor differentiation), followed by Class lb (change in vowel quality, i.e. greater differentiation) and Class lc (addition of a segment, i.e. complete differentiation). Level 2, similarly, shows its lowest rate of agreement in Class 2a (nasalization, i.e. minor differentiation), followed by Class 2b (addition of a syllable, i.e. greater differentiation) and Class 2c (desinences with no shared segment, i.e. complete differentiation).

Recall that, in addition to Class 2c, which shows total differentiation of the stressed desinences, we set up for purposes of verification a special Class 2d, consisting of the single opposition elsao. As regards oppositional saliency, this item also shows total differentiation of forms, and these forms are always stressed. The opposition of Class 2c does, however, possess the constant unstressed verbal root in both items (fal- in our example), while a constant root

76




of this sort is absent in Class 2d. Table 3 indicates that this circumstance has little or no effect, since all three models show nearly identical results for these two classes. Furthermore, the log likelihood test in the logistic model shows that the small difference found in that model is not significant even at the .5 level. It can thus be concluded that Class 2d is really not different in behavior from Class 2c, and that these two classes should be amalgamated.

Class 2e contains the stress-shifting preterits-which, parallel to Class 2c, show no shared segment in the desinences that mark the singular/plural opposition. Here, however, the singular shows root stress; and the singular desinence is either zero, as infez 'did (sg.)', or unstressed [i], as in disse 'said (sg.)' The plural desinence is stressed -eram (disseram 'said [pl.]',fizeram 'did [pl.]'), identical to Class 2b. Thus the opposition in this class is [i, -0/-eru]; and in some verbs the root vowel also changes. Despite these nuances, the fundamental similarity to Class 2c remains clear: there is complete differentiation of the desinences, and they are stressed in the plural.

Table 3 shows slightly more agreement for Class 2e than Class 2c in all three models; but in the logistic model, the log likelihood test reveals that the minute distinction found there (0.78 vs. 0.80) is not valid even at the .7 level. These negative results are, of course, partially caused by the fact that Class 2e has the fewest tokens of all the classes. Until more data are forthcoming on this class than those found in the approximately 115 hours of speech considered here, it must be concluded temporarily that Class 2e is very similar in behavior to Class 2c, and that the nuances distinguishing them are not important. From this point on, Class 2c will be understood to include both 2d and 2e. The results of this amalgamation are shown in Table 4.

CLASS EXAMPLE FREQUENCY PROB ADJ DEV la comelcomem 110/755 = 14.6% 0.11 -33.7% b falalfalam 763/2540 = 30.0% 0.26 -18.2% c fazlfazem 99/273 = 36.3% 0.35 -11.3%

2a ddldao 604/927 = 65.2% 0.68 + 17.9% b comeulcomeram 266/365 = 72.9% 0.78 +25.5% c faloulfalaram 1160/1450 = 80.0% 0.85 +33.6%

(Beta = 0.74.) TABLE 4.

Comparison of Tables 3-4, which refer to the same data under slightly dif- fering analyses, reveals several points of interest. The raw frequencies for all classes up to 2c do not, of course, change at all; but the new result for 2c represents an average of the former 2c, 2d, and 2e weighted by number of tokens. In the logistic model, however, all the new probabilities are numerically different from those of Table 3, although the relationships among various categories have not changed. The reason for this is the mathematical constraint imposed upon the average of the terms in Formula 5 for each group. Since the number of factors in the group has changed from eight to six, the averaging gives different results. We need not, however, concern ourselves with this matter, since all that is of interest is the constant relationship among categories. In the adjusted deviations, the new results bear the same sort of relationship

77




to the old as in the case of the raw frequencies. It is important to note, however, that the partial beta, which reflects the proportion of total variation accounted for by the factor group, remains constant at 0.74. Thus, confirming the log likelihood test in the logistic model, the more complex analysis gives no im- provement in the results of the additive model.

The final morphological categorization of Table 4 was subjected to two further types of test. First, using the log likelihood technique, it was confirmed that the results for these six categories are indeed statistically different at the .005 level; thus no additional amalgamations would be valid. Then several attempts to improve the analysis were made by splitting some of the six categories into subcategories. For example, in Class 2b the opposition foilforam, the irregular preterit of both ser 'to be' and ir 'to go', was separated from regular 2nd-3rd conjugation preterits. Forfoilforam a frequency of 111/156, or 71.2%, was found; the regular preterits' rate of agreement was 155/209, or 74.2%. In the logistic model, this gave results of 0.72 vs. 0.76, a distinction which the log likelihood test shows to be insignificant even at the .25 level. Based on this sort of statistical testing, I conclude that the six-leveled hierarchy realistically portrays the true situation.

5.1. A FEATURE ANALYSIS OF SALIENCY. In the informal discussion of the concept of oppositional saliency, it was argued that two main forces were simultaneously operating. The first of these was stress; and it was assumed, quite naturally, that stressed oppositions are more salient than unstressed oppositions. This distinction was used to set up two levels, on each of which the secondary criterion of material differentiation was used. Thus one might argue that the hierarchy of morphological categories is a derivative notion, and that saliency would be more insightfully analysed in terms of its component features.

To carry out such an analysis, two new factor groups were postulated- stress and differentiation:

(7) STRESS: Class a. unstressed (Morphological Classes la-c)

b. stressed (Morphological Classes 2a-c) MATERIAL DIFFERENTIATION:

Class a. minor differentiation (Morphological Classes la, 2a) b. greater differentiation (Morphological Classes lb, 2b) c. complete differentiation (Morphological Classes Ic, 2c)

Through appropriate recoding commands of SWAMINC4 and SPSS, the factor group representing the morphological hierarchy on the original data cards, was replaced by the appropriate values of these two factor groups. For example, the symbol representing Morphological Class lb was rewritten as symbols representing Stress Class a and Material Differentiation Class b. The results are shown in Table 5.

For the Stress feature group, all three models agree and show extremely sharp polarization, indicating that this is a very important determinant of agreement. In the Material Differentiation constraint group, however, we see for the first time a case where the raw frequencies do not agree with the probabilities

78




TYPE CLASS FREQUENCY PROB ADJ DEV

STRESS

a unstressed 972/3568 = 27.2% 0.22 -20.6% b stressed 2030/2742 = 74.0% 0.78 +26.8%

beta = 0.70 MATERIAL DIFFERENTIATION

a minor 714/1682 = 42.4% 0.35 -10.8% b greater 1029/2905 = 35.4% 0.54 +2.0% c complete 1259/1723 = 73.1% 0.62 +7.2%

beta = 0.20

TABLE 5.

of the logistic model, or with the adjusted deviations of the additive model. Both the latter are in accord with our expectations: increasing material differentiation correlates with increasing rate of agreement. Furthermore, although some of the numerical differences involved are relatively small, the two appropriate log likelihood tests show that the results are statistically distinct at the .005 level.

The reasons for the deceptive results in the raw frequency column are varied. In the first place, Differentiation Class a is a more or less evenly-weighted average of Morphological Classes la and 2a; but Differentiation Class b is heavily weighted toward Morphological Class lb, and away from 2b, because the former happens to have far more tokens. Just the opposite occurs with Differentiation Class c, which is heavily weighted in favor of Morphological Class 2c, and away from Ic. Thus Differentiation Class b is weighted in favor of the lower rate of agreement class, while Differentiation Class c is weighted toward the higher rate of agreement class. If all three classes had been evenly weighted, the results for Differentiation Classes a, b, and c would have been 39.9%, 51.5%, and 58.2%, respectively. It is just this sort of infelicitous distribution of data that was referred to in the first paragraph of ?4.2, above, as invalidating raw frequencies as a generally valid measure of effect. What is important is the joint effect of the stress and differentiation features: for each class of stress, each class of differentiation has a parallel effect, namely that of increasing (or decreasing) rate of agreement. The additive and logistic models are capable of reflecting this fact, despite the skewed distribution of data.

5.2. RESULTS FOR POSITIONAL CATEGORY. The ordering of the positional category was guided by the same central notion of saliency used in the morphological category. Here, however, the relevant type of saliency is not oppositional, but depends instead upon the relative position of the subject-which is the element that determines the appropriate form of the verb, and the verb itself.

The positional relationship is most salient when the determining subject immediately precedes the determined verb. This situation, in which the deter- miner/determined relationship is totally transparent, is labeled as Positional Class la, defining 'immediately' as allowing up to five syllables of intervening material. This cut-off point, while essentially arbitrary, has the result of allowing a series of two or three short adverbs (such as jd 'already', ndo 'not', or

79




nunca 'never'), while excluding most relative clauses and adverbial phrases. Less salient is Positional Class lb, in which more than five syllables separate subject from verb within the same surface sentence. The situation in which the subject/verb relationship is least salient is Positional Class Ic-where the determining element, the subject, follows the determined element, the verb.

The so-called 'hidden' or deleted subject presents a special problem. It is clearly distinct from any of the classes discussed above, because the subject is not even present in the surface sentence-the unit usually taken as basic in syntactic integration.?0 Thus the verbal desinence is not redundant within the sentence; and if the verb and its extra-sentential subject are to have any structural link at all, it can only be accomplished through the agency of agreement. Otherwise, there could be no grammatical algorithm to establish the cross- sentential connection from verb to subject, since there would be no mark whatever of plurality in the sentence containing the verb with deleted subject. On this functional basis, 'hidden' subjects are set up as a separate case outside the positional hierarchy, which is valid only within the limits of a single surface sentence." The positional classes are summarized below, with the positional hierarchy listed in descending order:

(8) CASE 1 (within a single surface sentence): Class a. immediately preposed subject (eles falam)

b. distant preposed subject (eles ... falam) c. postposed subject (falam eles)

CASE 2 (cross-sentential): Class a. deleted subject (... falam)

Results for the positional variable are shown in Table 6.

CLASS EXAMPLE FREQUENCY PROB ADJ DEV

la eles falam 1802/3680 = 49.0% 0.71 +3.3% b eles ... falam 84/254 = 33.1% 0.41 - 16.1% c falam eles 59/252 = 23.4% 0.24 -28.2%

2a ... falam 1057/2124 = 49.8% 0.65 -0.4%

(Beta = 0.21.) TABLE 6.

Within sentential limits (i.e. Case 1), all three models show parallel results that accord perfectly with the principle of saliency: Class la, consisting of immediately preposed subjects, shows the highest rate of agreement-followed by Class lb, where the subject is still realized and preposed, but separated from the verb by some distance. The lowest rate of agreement of the hierarchy is induced by the postverbal subject of Class Ic.

Case 2 uniformly shows results of approximately the same magnitude as Class la. I see no way in which this fact can be explained in terms of the

'0 For an explicit model along these lines, see Kimball 1973. " It would, in fact, be tempting simply to exclude sentences without realized surface subjects

from the data base, if this did not involve the loss of over 40% of the data. Nonetheless, the results obtained in this way do not differ markedly from those reported in the text.

80




principle of saliency; rather, it seems necessary, as noted above, to postulate that other principles of a functional nature interact with saliency, inhibiting what would amount to complete loss of any syntactic realization of plurality in a sentence with deleted subject and unmarked verb.

6. SOCIAL DIMENSIONS. All runs reported above contained, in addition to the structural constraints (morphological and positional factor groups, or the two groups constituting the feature analysis of the morphological classes as well as the positional group), a factor group consisting of identifications of each individual speaker. This was done in order to include in the calculations the social variables that had been left out of explicit consideration. In fact, the logistic model probabilities for these factors range from 0.20 to 0.89, and the adjusted deviations from -24.9% to +30.9% (partial beta = 0.43), showing that social variables must be quite important for a complete account of verbal agreement.

Table 7 reports the over-all frequency of application of the agreement rule for each speaker. The TOTAL column gives the number of verb tokens with plural subjects, while the APPLICATIONS column gives the subtotal of those verb tokens that actually showed a plural inflection. The table is in order of decreasing relative frequency, given as a percentage.

NAME APPLICATIONS TOTAL FREQUENCY

SON 421 523 80.5% VAN 180 258 69.8% CAN* 96 141 68.1% SID 312 505 61.8% UBI* 315 622 50.6% MAD* 132 263 50.2% ELV* 344 701 49.1% LID 72 184 39.1% HEN 127 328 38.7% ILD* 191 494 38.7% JOM 139 368 37.8% GCL* 77 209 36.8% FAU 299 818 36.5% IZE 145 417 34.8% MAR 114 346 33.0% PAU 28 91 30.8% NAT 10 42 23.8%

Total 3002 6310 47.6%

TABLE 7. (Asterisk marks older speakers.)

On the average, this sample of illiterate speakers applies the agreement rule in about half the appropriate cases; but there is an extremely wide range of variation, from NAT's 24%, on one extreme, to SON's 81%, on the other. As it happens, these two speakers are approximately the same age, and both grew up in the city of Rio. They are of opposite sexes; but MAR, near the bottom of the list, shows the same age group, area of origin, and sex as the speaker with the highest incidence of agreement. Similarly, SID has the same social characteristics as NAT, but is very near the top of the list.

81




The older speakers show considerably more uniformity of behavior than the younger, being concentrated in the upper range with an over-all variation of only about 25% (approximately 40%-65%). Nonetheless, younger speakers are found above, below, and among the older ones.

The general impression of indifference of effect of the social variables listed in Table 2 is reinforced by inspection of the general distribution of frequencies with respect to these categories, shown in Table 8.

WEIGHTING TOKENS SPEAKERS

city 1605/3216 = 49.9% 45.0% environs 1397/3094 = 45.2% 47.6%

younger 1847/3880 = 47.6% 44.2% older 1155/2430 = 47.5% 48.9% male 1176/2490 = 47.2% 43.1% female 1826/3820 = 47.8% 49.1%

TABLE 8.

The second column of Table 8 was calculated by simply aggregating all data across all speakers. This procedure has the effect of weighting more heavily the behavior of people who happen to have produced more data. For example, in the aggregated sample, the behavior of FAU, with 818 data tokens, is weighted about 20 times more than that of NAT, who has only 42 data tokens. In part, this sort of weighting is justified because the results for speakers with more data are probably more reliable; but good statistical practice would require a more sophisticated weighting algorithm. The third column of Table 8 goes to the opposite extreme, weighting each speaker equally, without taking into account quantity of data (or any other property) in any way.

To judge by the results of Table 8, the social factors reported there would not be thought very important. The largest differences are on the order of only about 5% ('city' ahead of 'environs' when weighted by number of data; 'female' ahead of 'male' when weighted by speaker). Furthermore, whatever weighting procedure one might choose would have to provide results intermediary between those of the second and third columns, reducing these small differences even further.

However, serious difficulties of two sorts invalidate the results of Table 8. In the first place, the data of 'Competencias Basicas do Portugues' were not controlled for a balanced distribution of social variables: the principal sponsor was, at the time, interested only in grammatical information that could be used in the planning of teaching materials and techniques. For example, of the nine speakers who spent their formative years in the city, six are males; but of the eight speakers brought up in the outlying areas, only three are males. In numerical terms, of the 3216 tokens produced by the city group, 1930 (or 61.0%) were from males; but of the 3094 tokens spoken by the group from the surrounding areas, only 560 (or 18.1%) were produced by males. This is exactly the sort of skewed distribution of data that can lead to deceptive raw frequencies, as noted above in ?5.1. Furthermore, there are even some combinations of social categories, such as older females from the city, that are not represented

82




at all in the data, creating empty cells. However, since both the logistic and the additive models are capable of dealing with skewed distributions and empty cells, these statistical difficulties can be overcome.

A second, and more fundamental, problem is that the proposed social variables include no measure of the speaker's possible relationships to other socio- economic levels in the surrounding community. As already mentioned, the speakers under study are all illiterate and from the same (low) socio-economic level. They are, further, all oriented toward improving their lot, a fact attested by their presence in adult literacy courses. Without exception, they hope to 'better' their lives through the course. One even dreams of becoming a doctor, while another hopes eventually to own a stable of race horses and an apartment in the most expensive area of the city. In general, older speakers have more modest ambitions. Thus ELV (age 45) hopes to 'have a family and get married' (she is a maid, living by herself in a room in her employer's apartment); UBI (age 42) wants only to 'stop being a janitor' (his current profession); ILD (age 43) hopes to become a seamstress or a manicurist, etc. By way of contrast, SID (age 16), HEN (age 17), and MAR (age 26) want to travel outside Brazil as a preliminary to their future careers, a privilege usually reserved for the children of the elite. VAN (age 15) is the would-be doctor, while IZE (age 23) hopes to get a banking job. Although a study of such aspirations would indeed be interesting, the high correlation with age rules out their use as a social indicator.

What we really wish to measure is the degree to which a given speaker has managed to penetrate the dominant culture of the middle and upper-middle socio-economic levels.12 I have been careful to limit the desired measure to middle-level CULTURE because, in point of fact, only one of the 17 speakers considered here has anything like middle-class status in real life.'3 Despite this lack of actual status, however, some speakers show that they are able to participate in middle-class concerns and anxieties, and seem able to relate to their middle-class interviewers perfectly well on this level. These speakers have, as it were, a clear view from the outside, although they are not actually able to enter the higher socio-economic levels; the social view of the others is strictly limited to their own level.

Since the interviews were conducted with no specific concern for measuring social factors, we must be satisfied with an indirect indicator. For this purpose, I propose to use the speaker's reaction to television. In this connection, it is important to bear in mind that the target of Brazilian television programming seems very narrowly confined to the urban middle (or upper-middle) class- probably because this group, although numerically small in comparison to the total population of the country, controls a very large proportion of the total purchasing power. Prime evening time is taken up mainly by 'novelas', i.e.

12 My occasional use of the term 'class' is strictly non-technical. In particular, I do NOT have

in mind the Marxist concept of class or any of the associated theoretical apparatus. 13 This is CAN, who lives in his own apartment in one of the less expensive middle-class areas

of the city. All the other speakers either live with their employers or in areas of markedly lower socio-economic status.

83




short-run soap operas that are heavily based on middle-class morality and anxieties. These soap operas are followed by their viewers with great dedi- cation, and are successfully used by businessmen as a basis for a whole series of secondary products, ranging from record albums and magazines through jewelry and other fashion items. Most of the comedy shows, too, presuppose a cultural and political background not attained by the people in our sample. Nonetheless, a relatively smaller proportion of broadcast time (mainly Saturday and Sunday afternoons) is given over to variety shows featuring popular sing- ers, sambas, amateur competitions of various sorts, and so on.

During investigations preliminary to the 'Competencias Basicas do Por- tugues' research project, it was quickly discovered (contrary to expectation) that most of the speakers interviewed watched little if any television, even when they had free access to a set. Those who did watch television overwhelmingly preferred the weekend variety shows or broadcasts of soccer matches. When requested, most of the illiterate people interviewed were unable to recount the plots (or even, for that matter, the titles) of the most popular novelas. The reason for these people's lack of interest in novelas seems to be their inability to follow the story lines: they simply cannot understand the motivation behind the characters' actions and reactions.

In the sample of 17 speakers under study here, only five are followers of novelas. Of these, four are from the younger group-two females (SON, VAN) and two males (SID, JOM)-while one (CAN, a male) is from the older group. The remaining twelve speakers either watch no television at all, or watch mainly the variety shows. Table 9 shows the usage of the verbal agreement rule by these two groups. In accord with the ideas developed above, novela- watching is taken as indicative of a wider cultural view-or what I shall call, at Gillian Sankoffs suggestion (p.c.), a 'vicarious' orientation, in contrast to the 'experiential' orientation of the remaining speakers.

WEIGHTING TOKENS SPEAKERS

vicarious 1148/1795 = 64.0% 64.9% experiential 1854/4515 = 41.1% 42.1%

TABLE 9.

Although based only on potentially-deceptive raw frequencies, Table 9 confirms the idea that cultural orientation is an important determinant of rate of use of the agreement rule. However, given the highly skewed and partially defective distribution of data within the various social and linguistic categories, use of a statistical model of joint effect becomes imperative.

In the logistic model, D. Sankoffs VARBRUL2 program imposes no limi- tations on the quantity of data or factor groups in a particular analysis, aside from the storage capacity of the computer utilized. However, SPSS (version 7) limits the capacity of its MCA analysis to only five factor groups per analysis. This limitation constitutes a serious problem, since we have four candidates for social groups (orientation, sex, age, origin) and three for structural groups (stress, differentiation, position), making a total of seven. For this reason, the data, analysed according to all seven factor groups, were first run in VAR-

84




BRUL2 (see Table 10 for results), and appropriate log likelihood tests were performed in order to evaluate the statistical significance of each group. These tests show that each of the postulated factor groups, with the exception of origin (city of Rio or surrounding areas), is significant at the .005 level or better, but that this latter group's significance is worse than .75. The tests ranked the factor groups in the following order: stress, cultural orientation, position of subject, material differentiation, age, sex. The first five of these groups were then run in SPSS's MCA program. In order to test the remaining groups (sex and origin), the stress and material differentiation factor groups were replaced by the morphological categories of Table 4, and two separate runs were made, each with the four surviving factor groups (morphological category, cultural orientation, position of subject, age) as well as one of the two remaining variables. For this reason, the additive model results in Table 10 are, in fact, from three separate runs.

TYPE CLASS

STRESS

a unstressed b stressed

CULTURAL ORIENTATION

a vicarious b experiential

FREQUENCY PROB ADJ DEV

972/3568 = 27.2% 2030/2742 = 74.0%

1148/1795 = 64.0% 1854/4515 = 41.1%

0.24 0.77

0.69 0.31

-20.5% +26.7%

beta = 0.69

+ 19.7% -7.8%

beta = 0.37 POSITION OF SUBJECT

la preposed b distant c postposed

2a deleted

1802/3680 = 84/254 = 59/252 =

1057/2124 =

49.0% 33.1% 23.4% 49.8%

0.71 0.41 0.24 0.65

+3.6% -17.6% -29.9% +0.5%

beta = 0.22 MATERIAL DIFFERENTIATION

a minor b greater c complete

714/1682 = 1029/2905 = 1259/1723 =

42.4% 35.4% 73.1%

1155/2430 = 47.5% 1847/3880 = 47.6%

1826/3820 = 47.8% 1176/2490 = 47.2%

1605/3216 = 49.9% 1397/3094 = 45.2%

0.34 0.54 0.61

0.58 0.42

0.54 0.46

0.50 0.50

-11.0% +2.3% +6.9%

beta = 0.20

+6.9% -4.3%

beta = 0.16

+2.1% -3.3%

beta = 0.08

+1.6% -2.9%

beta = 0.06

(r = 0.82.) TABLE 10.

The first data column of Table 10 repeats information on raw frequencies that has already been presented for individual factor groups. The second (lo-

AGE

SEX

ORIGIN

a older b younger

a female b male

a city b environs

85




gistic model) and third (additive model) columns, however, represent for the first time a complete overview of the joint effect of all variable constraints, both structural and social. The results for the former differ very little from those already presented, since the social factors had been included implicitly in the earlier calculations by means of variables representing individual speakers. The various factor groups are listed in Table 10 in order of statistical significance within the logistic model as determined by the log likelihood test. This corresponds exactly with the ordering according to partial betas.

The statistic r is, roughly speaking, to the whole what the betas are to the parts-it indicates (exactly), when squared, the total proportion of variation in frequency of agreement attributable to the additive effects of all the factors. The level of better than 0.8 achieved here is generally considered extremely successful in empirical statistical investigations.

The results for the social variables are quite revealing. By far the most important social conditioning is attributable to the speaker's degree of pene- tration into the culture of the surrounding higher socio-economic levels, as reflected in his or her habits of television viewing. It is true, of course, that television provides a massive dosage of the standard language with its categorical agreement rule; but the same is true of radio, a medium directed more toward the lower classes and enjoyed by a good number of the speakers in our sample-even if its horoscopes, advice to the lovelorn, outraged commentary on local issues etc. are interspersed with a good deal of popular music. The devotion to novelas on the part of these people would seem to indicate that their comprehension of middle-class life style has reached the point where they can follow the plots and develop an interest in them. It is the fact that these speakers participate vicariously in a socio-cultural context foreign to their own milieu that is reflected by the cultural orientation variable.'4

The age variable, which seems indifferent in terms of raw frequencies (but see the last column of Table 8, where the data from each speaker are weighted equally), shows a significantly higher rate of agreement for the older group when the simultaneous effects of the other variables are taken into account on either the logistic or the additive model. This age-grading in apparent time is the 'synchronic' reflex of the 'diachronic' trajectory of the agreement rule from the grammar. As we have come to expect in this sort of linguistic change, women are closer to the prestige form than men-although, in this particular case, the difference is very small and accounts for only a minute proportion of the over-all variation, a fact reflected in an extremely low partial beta. The area of origin of the speakers shows no effect at all in the logistic model, and a negligibly small effect in the additive model.

14 For the record, I know of one person for whom this variable does not provide the desired results. This speaker, diagnosed as mentally retarded by the state authorities, has an over-all rate of application of the agreement rule of only about 30%; yet she spends virtually the whole day planted in front of a television set and thus watches the full range of programming, from the variety shows through the novelas and even news broadcasts. Unfortunately, a thorough investigation of the assumed general correlation between television viewing habits and cultural orientation would require sophisticated sociological and anthropological techniques far beyond the goals of this paper.

86



DIMENSIONS OF A SYNTACTIC CHANGF

In the social variables, as with the material differentiation variable, there is a certain lack of correspondence between raw frequencies and the results according to the models of joint effect. As already noted in the discussion of Table 8, this results in part from the skewed distribution of data among speakers. In fact, both models provide results more similar to the last column of Table 8 than to the 'Tokens' column. There is, furthermore, a significant pe- culiarity in the behavior of the cultural orientation and age variables. Consider the frequencies reported in Table 11.

OLDER YOUNGER

All 47.5% 47.6% Vicarious 68.1% 63.6% Experiential 46.3% 35.7% TABLE 11. Raw percentages of verbal agreement

according to age and cultural orientation.

The first row of this table repeats the over-all frequency for all older and younger speakers, considered as an aggregated group; the second and third rows break down this group into vicarious and experiential orientation subgroups. Within both age groups, the speakers with a vicarious cultural orientation show more agreement than those with an experiential orientation; within each orientation group, older speakers agree more than younger speakers. Unfortunately, this latter fact does not come through to the over-all figures for age groups, because the skewed distribution of data causes the weighted averages in the first line to turn out nearly equal. Nonetheless, the models of joint effect are capable of overcoming this difficulty; they present a correct evaluation of the effects of each variable.

Note that the orientation variable has a much greater effect than the age variable. This can be seen in several ways. Formally, the partial beta of the additive model is more than twice as large for orientation as for age. Further- more, the net difference associated with the vicarious/experiential dichotomy is 27.5%, while that of older/younger is only 11.2%. In the logistic model, the probabilities for the orientation categories are more distant from 0.50 (the neutral point) than those of the age variable. Finally, informal inspection of Table 11 quickly makes one realize that the differences between the rows are far greater than those between the columns.

In sociolinguistic terms, the relative strength of these two variable constraints reflects a real-world conflict between opposing trends. On the one hand, we see in the age variable the continuation of a slow process of rule death, related to similar changes in the NP and elsewhere. This variable is, so to speak, exterior to the speaker, impinging upon him or her without any internal motivation. The orientation variable, on the other hand, reflects a resurgence in the use of this same rule that comes from within the speaker, depending upon his or her own cultural orientation. In contrast to the relatively small effect of being older, possessing a vicarious cultural view implies a large differential in incidence of the agreement rule, when compared with other speakers having otherwise similar social characteristics.

87




It is tempting to speculate on what the long-term linguistic outcome of these opposing social forces might be. Purely on the basis of informal observation, I have the impression that the vicarious orientation is much more prevalent among the young than among the old; and the distribution in the sample under study here confirms this. If the vicarious orientation continues to spread among younger speakers, given the relatively small age-differential, the situation could soon reach a point where that group might produce a higher rate of agreement than the older speakers. Since there are no social barriers between the two orientation subgroups, at a certain point in time the resurgence in use of the agreement rule might even spread throughout the younger group, independently of the orientation variable, and produce a reversal in the trend toward the elimination of the rule. Future studies will determine if there is any validity to this speculation.

7. THE INDIVIDUAL WITHIN THE GROUP. All results presented up to now have referred to the behavior of a group of speakers defined in terms of social parameters. Given current psycholinguistic theories, according to which linguistic activity is controlled by the individual brain, such collective data would seem to be meaningful only insofar as all members of the group behave similarly; otherwise, group data would amount to no more than a meaningless average of disparate behaviors. If this occurred, the group figures would not describe correctly the performance of any individual, and could have no psy- chological reality. Furthermore, given that the forms used by a speaker are similar or identical in all relevant aspects to those used by other speakers, the scale of saliency is invariant across speakers, and there is therefore no reason to expect conflicting orderings of constraints in the hierarchized dimensions, either among speakers, or between individuals and the group.

To evaluate the behavior of the individual with respect to the group, I shall use techniques developed by Guy 1975, modified somewhat for the case under study here and incorporating recent statistical advances. In any comparison of statistics for a group of speakers with the corresponding results for its component members, it is very important to bear in mind the fact that the group always has the advantage of possessing a greater quantity of data; thus accuracy is increased, and the chances of noticeable fluctuations in less frequently occurring categories, resulting from probabilistically determined out- comes, are reduced. In this sense, the idea that the group is more regular than the individual is a triviality.

The evaluation of group vs. individual behavior will be carried out first for the morphological hierarchy, since this is the most complex constraint group and thus most likely to show distinctions. There are six morphological categories in the hierarchy. In principle, it would be possible for an individual speaker's results for a given category to fall anywhere with respect to any other category; i.e., Class la might turn out to show a higher incidence of agreement than Class 2c. In practice, however, we rarely find deviations of more than one rank with respect to the group pattern. Thus we are interested in determining the individual's behavior in five pairs of neighboring categories of the hierarchy. Earlier, in dealing with the group data (?5), the .005 level

88




was used as a criterion for accepting numerically distinct results for such neighboring categories as statistically valid. However, because of the relative scarcity of data on individuals, it becomes necessary to relax this criterion here to the 0.1 level.

For purposes of comparison with the grouped results, six individuals, representing a balanced distribution of the three most important social variables, were chosen. It was required that four of these speakers be from the younger group, and two from the older. Within each age sample, half were male, and half female. Furthermore, in the case of the younger speakers, one individual of each sex showed the vicarious cultural orientation, and one the experiential orientation.15 To ameliorate the problem of scarcity of data on individuals, the speakers actually selected in accord with the above criteria were those who had the greatest number of data. Because of low cell counts, however, some speakers showed categorical behavior (i.e. 0% or 100% rule application) in certain cells (e.g., IZE had 0% for Morphological Classes la and lc). The logistic model is unable to deal with such factors because the terms in Formula 3, of the form in 5, become zero or infinity under these conditions. For this reason, speakers with categorical cells were eliminated from the comparison. As it happens, the five individuals with the greatest number of data are included in the sample chosen in accord with the above requirements.

Table 12 reports the results for the morphological hierarchy in individuals.

CLASS FREQUENCY

Speaker: ELV la 9/92 = 9.8% b 102/297 = 34.3% c 20/32 = 62.5%

2a 56/91 = 61.5% b 25/37 = 67.6% c 132/152 = 86.8%

Speaker: FAU la 1/62 = 1.6% b 50/380 = 13.2% c 21/48 = 43.8%

2a 46/89 = 51.7% b 29/42 = 69.1% c 152/197 = 77.2%

Speaker: SON la 26/53 = 49.1% b 175/223 = 78.5% c 1/5 = 20.0%

2a 55/62 = 88.7% b 34/36 = 94.4% c 130/144 = 90.3%

PROB

0.08 0.29 0.53 = 0.57 = 0.73 0.87

0.03 0.22 0.60 = 0.68 0.83 = 0.89

0.23 0.54 * 0.06 0.70 = 0.85 = 0.76

TABLE 12.

FREQUENCY

Speaker: UBI 12/115 = 10.4% 89/219 = 40.6%

10/41 = 24.4% 86/107 = 80.4% 25/30 = 83.3%

93/110 = 84.6%

Speaker: HEN 5/36 = 13.9%

31/144 = 21.5% 3/17 = 17.7%

25/54 = 46.3% 14/18 = 77.8% 49/59 = 83.1%

Speaker: SID 20/59 = 33.9%

107/210 = 51.0% 10/15 = 66.7% 66/86 = 76.7% 35/39 = 89.7% 74/96 = 77.1%

15 It was not possible to impose this requirement on the older sample, since only one older speaker had vicarious orientation (CAN), and he provided too few data tokens to'make the comparison worthwhile.

PROB

0.08 0.35 * 0.18 0.78 = 0.82 = 0.86

0.16 = 0.24 = 0.19 0.51 0.90 = 0.88

0.18 0.31 = 0.44 = 0.62 0.82 * 0.63

89




Both raw frequencies and the logistic model probabilities are shown. The equals sign following a probability indicates that the difference between it and the next probability in the hierarchy (listed immediately below) is not statistically valid at the 0.1 level, a situation that I shall refer to as 'neutralized'. An asterisk indicates that the difference between categories is valid at 0.1 or a better level, but is the opposite of what would be expected from group behavior. No notation is made when the difference between probabilities associated with a class and the next class in the hierarchy is statistically valid and agrees with the group order.

In general, this table reveals that the behavior of individuals is in accord with the group pattern, but there are exceptions. Thus UBI's value for Class Ic is lower than for lb, and the same is true of SON; while SID's Classes 2b and 2c are in the wrong order. Of the thirteen neutralized pairs, ten are in accord with the group order, while only three are in the opposite order. Thus, of the thirty neighboring pairs in Table 10, fourteen are correct, thirteen are neutralized (but still about 80% in the correct order), and only three are in the wrong order. In the entire table, only one result is more than one rank away from the group order: this is SON's value of 0.06 for Class Ic, which is not only lower than her value of 0.54 for Class lb, but also lower than her value of 0.23 for Class la. Significantly, it is precisely SON's Class Ic that has the fewest data tokens in the table (only five).

It is instructive to consider further the question of number of data tokens- although this is not, of course, the only determinant of statistical significance. Since a comparison of values for members of a pair is only as accurate as the LEAST accurate of the values involved, the number of tokens in the SMALLEST cell of each pair is the number of empirical data tokens upon which the comparison rests. For example, the pair la-lb in ELV is taken as based upon 92 tokens (the total for la) rather than 297 tokens (the total for lb). Scoring in this manner, the fourteen correct pairs average 47.1 tokens, the thirteen neutralized pairs, 30.2 tokens; and the three incorrect pairs, 28.3 tokens. Fur- thermore, the results for the positional hierarchy are exactly parallel: of the twelve neighboring pairs of Case 1, eight are correct and show an average of 20.5 tokens, while four are neutralized (but in the correct order) and average only 14.0 tokens. As these figures show, it is not possible to establish a fixed requirement for statistical significance in terms of number of tokens. The principal reason is that the degree of difference of behavior between categories is also a strong determinant of statistical significance. We can conclude that apparent individual deviations from the group pattern tend to occur when the number of tokens is small, relative to the number of tokens in other cells- and that, for the most part, individuals do conform to the group pattern.-

8. THE ORIGIN OF LOSS OF VERB PHRASE AGREEMENT. Loss of agreement in the VP involves changes of very disparate sorts on the surface level. In the classes on the upper end of the morphological hierarchy, the process appears to be strictly one of generalization of privilege of occurrence of the singular morphemes at the expense of their plural equivalents (thus, in Class 2c, [-6]

90




seems to be taking over the territory of [-arf]). But on the lower end of the morphological scale, this same process appears to be the result of a simple phonological rule that denasalizes final vowels (e.g., in Class la, [i] becomes [i]). In fact, such a rule exists in Brazilian Portuguese and applies variably to classes other than verbs (i.e. to nouns such as vagem 'string beans', adverbs such as ontem 'yesterday', etc.) Furthermore, this rule is attested even in the oldest texts, and thus predates the evolution of a non-agreeing system by at least several centuries. In some areas of Portugal, principally in the region of Entre-Douro-e-Minho, the rule survives in popular speech, where its effects have been reported in both nouns and verbs of Class la (virgem 'virgin' as birge; vertem 'they spill' as berte-Vasconcellos [1901] 1970:86-7), although no general variability in agreement is mentioned for the areas in question. However, the view of loss of agreement as a consequence of a phonological change is fully applicable only in Class la; from Class lb upward, more than the simple loss of nasalization is involved, and the process can no longer be viewed as a phonological rule.

The developments in real time mentioned above lead one to investigate the corresponding phenomena in apparent time (i.e. in age groups). For this purpose, Table 13 lists, for all three models, the results for Class la in the older and younger groups, and compares these results with Class lb, the next category. The results of the logistic model for these two categories are shown to be statistically significant at the .005 level by the log likelihood test.

GROUP: OLDER GROUP: YOUNGER

MODEL CLASS VALUE DIFFERENCE VALUE DIFFERENCE

Frequency la 11.4% 17.0% b 31.4% 29.3%

20.0% 12.3% Logistic la 0.10 0.11

b 0.31 0.23 0.21 0.12

Additive la -36.8% - 31.0% b - 16.6% - 19.2%

20.2% 11.8%

TABLE 13.

In all models, the older speakers evidence a wider gap between these two classes; this indicates that, at earlier stages of the history of loss of agreement, Class la was even further in the forefront of non-agreement than it is now. Since it is precisely Class la that represents the uniquely phonological component of the change, this in turn suggests that the loss of final nasalization was once responsible for a larger proportion of the over-all process, and may have been an actuating force behind it.

If we look at the same problem from a different perspective, it is instructive to compare loss of nasalization in verbs with that in other grammatical categories. In verbs, several distinct subcategories of nasalization loss must be considered. The results of denasalization are summarized in Table 14.

91




CLASS FORM [ + NASAL] [- NASAL]

1 comem [k6mi] [k6mi] (= sg.) falam [fala] [fala] (= sg.)

2 falam [falu] [falu] (sg. fala) fazem [fazi] [fazi] (sg. faz)

3 comeram [kom6rfi] [kom6ru] (sg. comeu) falaram [falarfi] [falaru] (sg. falou)

TABLE 14.

In Class 1, the forms of the [ + nasal] column are plural, while those of the [- nasal] column are singular; here the loss of nasalization totally destroys the opposition. In Class 2, the forms of both columns must be classified as plurals, since they are both distinct from the singular, and the loss of nasalization does not destroy the opposition. Finally, in Class 3, as in Class 2, both columns are plural.

Votre 1978, 1979 has studied the general process of denasalization-in particular, the relationship of the rates of retention of nasalization in the three verbal classes of Table 14 vs. that in nouns. In these considerations, the nouns serve as a sort of control against which the behavior of the verbal classes is measured.

Table 15 shows Votre's results (1979) for retention of nasalization in two age

MODEL CLASS OLDER YOUNGER

Logistic Verbs 1 0.24 0.17 2 0.83 0.88 3 0.67 0.43

Nouns 0.24 0.46

Frequency Verbs 1 108/668 = 16.2% 31/202 = 15.3% 2 155/258 = 60.1% 24/44 = 54.5% 3 110/301 = 36.5% 10/99 = 10.1%

Nouns 454/683 = 66.5% 244/359 = 68.0%

TABLE 15.

groups, corresponding to subsets of the same speakers studied in the main part of this paper. Consider first the older speakers. In terms of logistic model probabilities, the four classes of Table 15 show a clear split into two levels- a high rate of retention of the nasal in verbs of Classes 2 and 3, and a low rate in Class 1 verbs and nouns. As far as the verbal classes are concerned, the relatively frequent absence of the nasal in Class 1, where it constitutes the plural desinence, is in accord with the principle that 'deletion of a morphemic segment is more likely ... if the morpheme is redundant' (Naro 1980:166). In this case, the plural desinence is most often determined by a previously occurring plural subject within the context of the discourse. The nasals in the other two verbal classes are not redundant, and are thus more resistant. The relatively low rate of nasal retention in nouns, however, is specific to the history of Portuguese, which has a general long-term drift toward variability in this class on both sides of the Atlantic. The most important point to note,

92




however, is the fact that, in the older speakers, loss of nasalization in Class 1 verbs, which is equivalent to loss of subject/verb agreement, is no more likely than loss of nasalization in nouns.

There are some striking inconsistencies between the logistic model probabilities and the corresponding raw frequencies in Table 15. In particular, although Class 1 verbs and nouns show equal probabilities of nasal retention, their frequencies for this same phenomenon are at opposite extremes, with Class 1 verbs at 16.2% and nouns at 66.5%. The reason behind this seemingly strange situation is the concurrent effect of stress. In verb Class 1, the nasal is always in an unstressed syllable, while in nouns it is often in a stressed syllable (alguem 'someone', mafa 'apple', etc.) This is a very important distri- butional property, since stressed nasals are nearly always retained (frequency 434/444 = 97.1%, probability 0.95), while unstressed nasals often disappear (frequency 393/1466 = 26.8%, probability 0.05). Thus the difference in frequencies between these two columns in the older speakers is caused by a strong correlation with stress, rather than by a loss of nasalization peculiar to one or the other class.

In the younger speakers, the situation is quite different. For this group, the frequencies and probabilities agree in attributing a considerably lower rate of nasal retention to verb Class 1 than to nouns. Furthermore, verb Class 3, which had a higher rate of retention than nouns for the older speakers, is equal to Class 1 for the younger speakers. This means that in the younger group, according to the logistic model, verb Class 1 does have its own unique loss of nasalization. In other words, in this age group, the loss of nasalization in Class 1 verbs, and consequently of subject/verb agreement, can no longer be attrib- uted to a correlation with stress, even though stress remains powerful (in the logistic model, 0.93 for stressed and 0.07 for unstressed). Similarly, the peculiar loss of nasalization in verbs has spread, although not as strongly, to verb Class 3, where nasalization is not redundant. Verb Class 2, however, has not yet been affected. From these facts we may conclude that, in the younger group, the loss of nasalization in Class 1 verbs has at least two components-a general denasalization that can be seen in the nouns, and a specific morphological shift in favor of singular desinences. The over-all view that emerges is this: within the class of verbs, the phonological process of denasalization (and consequent loss of agreement) made its first inroads at the weakest point (i.e. Class 1, where the nasalization is syntactically determined), later generalizing to other environments.

In general terms, the evidence indicates that, in the earlier stages of the evolution of verbal agreement, environments involving simple loss of nasalization were further in the forefront of agreement reduction than in later stages- and that, in the earlier stages, nasalization loss in Class 1 verbs was more similar (in fact, nearly equal) to nasalization loss in non-verbs. Extrapolating backward, I conclude that one of the actuating forces that set off the evolution toward a non-agreeing morphological system was the phonological rule of denasalization. This rule was, and continues to be, variable.

93



LANGUAGE, VOLUME 57, NUMBER I (1981)

9. SALIENCY HIERARCHIES IN NOUN PHRASE AGREEMENT. As in the VP, the

agreement rule in the NP is disappearing in modern Brazilian Portuguese. However, this change has a wider socio-economic distribution, and seems to be further evolved. Within the NP, variation in the realization of plural morphemes is extensive, even in highly educated speakers who show nearly categorical marking of the verb.

The behavior of nominal plurals has been studied in the area of Minas Gerais by Braga 1977, and in the area of Rio de Janeiro by Scherre 1978. These two investigations produced essentially parallel results; but since the latter work is partially based on a subgroup of the speakers in our sample, it will be cited here. Scherre's illiterate speakers were all selected from among the younger group. The over-all frequencies are given in Table 16 (Scherre. p. 98).

NAME APPLICATIONS TOTAL FREQUENCY

VAN 1084 1588 68.3% SID 923 1365 67.6% SON 619 947 65.4% HEN 518 992 52.2% IZE 587 1144 51.3% JOM 738 1559 47.3%

TABLE 16.

In addition, Scherre presents data from four speakers of the same age level, but of higher socio-economic and educational status. These speakers average a frequency of 80.0% in application of agreement.

A hierarchy of saliency of the singular/plural opposition, similar to that of Table 3, was set up for the nominals and postulated as a constraint group. Pairs in which the plural mark is in an unstressed syllable were assigned to Level 1, while Level 2 was reserved for the more salient case in which a plural mark occurs in the stressed syllable. Within Level 1, two classes were distinguished. In Class la, the plural differs from the singular merely by the addition of a final -s (ano/anos 'year/years'), but in Class lb the plural form acquires an extra syllable through the addition of-es (lugar/lugares 'place/places').16 Level 2 also has two subclasses. Class 2a consists of pairs in which the plural -s attaches to the stressed final syllable, usually with further morphophonemically determined changes: ancestral/ancestrais [-aw/-ays] 'ancestor/ancestors', dis- cussdo/discuss6es [-iaw/-6ys] 'discussion/discussions'. At the top of the hierarchy is Class 2b, containing the so-called metaphonic plurals, in which there is a very salient change in the quality of the stressed root vowel with concurrent addition of -s. (ovolovos [dvu]/[5vus] 'egg/eggs'). A summary of the hierarchy of oppositional saliency is as follows:'7

16 Items of the type vez/vezes 'time/times', in which the -es follows another syllable ending in a sibilant, were not included in Class Ib: they undergo a haplological process of reduction, independent of the general agreement rule.

17 The final -s of Classes lb, 2a, and 2b may be deleted. In Scherre's data, the correspondence to stress is not perfect, since she includes plurals of the type cafes 'coffees' in Class la and of the type fdceis 'easy' (pl.) in Class 2a. Nonetheless, these items occur infrequently.

94




(9) LEVEL 1 (unstressed): Class a. [-k/-s]

b. [-0/-es] LEVEL 2 (stressed):

Class a. [-aw/-ais, -ew/-eis, -iw/-is, -aw/-6os, -aw/-as, -aw-aws] b. [6/3-s]

The results obtained by Scherre (p. 79) for the morphological variable in the two socio-economic groups she studied are represented in Table 17.

CLASS EXAMPLE FREQUENCY PROB

Educated speakers la ano/anos 2705/3365 = 80.4% 0.39 b lugar/lugares 76/86 = 88.4% 0.57

2a discussaoldiscussoes 116/122 = 95.1% 0.85 b ovo/ovos 21/21 = 100.0% +

Illiterate speakers la anolanos 4291/7240 = 59.3% 0.29 b lugar/lugares 37/80 = 46.3% 0.25

2a discussaoldiscussoes 39/101 = 38.6% 0.49 b ovolovos 30/34 = 88.2% 0.96

TABLE 17.

Considering first the educated group, we see a perfect correlation between increasing oppositional saliency, on the one hand, and increasing frequency or probability of actualization of a plural mark, on the other. For the illiterate group, the logistic model probabilities reveal a neutralization of effect on Level 1 (unstressed), but this level can be clearly distinguished from Level 2 (stressed), which is not neutralized.

As regards the morphological constraint group of the illiterate speakers, there is a discrepancy between the relative orderings of the probabilities and the frequencies. This is caused by the distribution of these constraints with respect to the positional group. In NP's, the first element, labeled as position 0, nearly always carries an actualized plural mark, while in succeeding positions the chances of plural marking decrease, as shown in Table 18.

POSITION FREQUENCY PROB

0 3212/3289 = 97.7% 0.98 1 1172/3847 = 30.5% 0.45 2 84/416 = 20.2% 0.31 3 1/39 = 2.6% 0.06 4 4 = 0.0%

TABLE 18.

These results for the nominal position group are important for my purposes, since they confirm the operation of the principle of saliency along the same lines as in the verbal positional group. The interaction with the morphological group results from the fact that position 0, in which marking is nearly categorical, is almost always occupied by an article (os, as 'the') or a demonstrative (estes 'these', aqueles 'those'), both of which belong to Morphological Class

95




la. Thus the raw rate of agreement for Class la is inflated because a relatively higher proportion of la tokens occurs in position 0 than tokens of the other morphological categories.

The results of Table 17 suggest that there is no relationship between frequency of occurrence of a paradigmatic class and the corresponding rate of agreement. In the educated group, for example, the most frequent class (la, with 3365 tokens) has the lowest rate of agreement, while the least frequent class (2b, with only 21 tokens) shows categorical agreement in the sample. The situation is parallel in the case of VP agreement-referring back to Table 3, the highest rate of agreement is found in the least frequent class (2e, with 116 tokens), while the most frequent class (Ib, with 2540 tokens) shows the second lowest rate of agreement. It might seem, however, that in considerations involving frequency, the appropriate unit would be the individual morphological item rather than paradigmatic classes. Indeed, the item ser 'to be' (elsdo 'is/ are', Class 2d of Table 3) has a high frequency of occurrence and a high rate of agreement, while the individual verbs that compose the less frequently agreeing classes would each also have a low frequency. Note, however, that Class 2e has a low frequency but a rate of agreement that exceeds even that of Class 2d. Furthermore, the subclass of 2e consisting exclusively of the items disseldisseram 'said (sg./pl.)' and trouxeltrouxeram 'brought (sg./pl.)' (combined with Class 2e in this paper precisely because of low frequency; see fn. 8) shows agreement in 28 of the 31 (= 90.3%) instances in the data. Thus, even though these two morphological units occur very infrequently, each has a higher rate of agreement than the very frequent elsdo. Although I do not have data for individual items on the lower extreme of the hierarchy, the results so far available make it seem unlikely that further investigation would reveal a correlation between frequency of occurrence and rate of agreement on either the paradigmatic or individual item levels.

10. CONCLUSION: THE ROLE OF SALIENCY IN SYNTACTIC CHANGE. Two aspects of the over-all process of linguistic change may be distinguished for analytical purposes: (a) ACTUATION, i.e. the beginning point, or first context, of a change; and (b) DIFFUSION, i.e. the subsequent spread of the change to other environments.

In the loss of verbal concord, I have argued above that the linguistic actuating force was a low-level phonetic rule of denasalization of final vowels. For the verbs of Morphological Class la, this rule, operating upon a plural form such as comem [komi], produces a form that coincides exactly with the singular.18 Thus, for speakers having this rule, the singular/plural opposition collapses on the surface in a certain part of the language. The diffusion of the non-agreeing system throughout the language then proceeds, in accord with the principle of saliency, along several distinct dimensions-extending itself most strongly in those contexts where it involves the smallest (i.e. least noticeable) change in surface form. The change thus sets in at the zero point of surface differentiation between the old and the new systems, and spreads to other points along the

'8 This is also partially true of Level lb, where some speakers may show a [fala] / [fala] opposition rather than the more usual [fala] / [falu].

96




path of least surface differentiation. This view of the process reinforces and confirms the general idea of syntactic change as a surface phenomenon (Naro 1976), according to which change originates in contexts that admit more than one syntactic analysis with no change in surface form.

Since salience is a general property of the linguistic system, one would not normally expect groups of speakers at different stages of evolution to exhibit conflicting orderings of variable constraints in the hierarchized dimensions, although such constraints can be neutralized under certain circumstances. The principal difference between a highly advanced and a less advanced group of speakers should lie in the general propensity to use the rule. In other words, as noted in ?7, the relative impedance of the hierarchized constraints should be the same across speakers, because the forms involved-and thus their relative salience-are the same. However, the general level of use of the rule might vary from group to group without altering the relative ordering of the constraints. In fact, this is exactly what we have found to be the case in the study of agreement in the NP and VP.

Thus we conclude with a general view of a major type of syntactic change as originating in an environment, or set of environments, where the structure of the rest of the language is such that more than one construal (i.e. syntactic derivation) of the same surface structure is possible. However, even though this particular originating environment may involve no (or only very incon- spicuous) surface differences between the old and new analyses, consequences in other parts of the language will lead to the creation of forms (or whole new construction types) not previously present. These new items do not, in general, enter the language simultaneously, or even at the same rate-despite the fact that, from the linguist's point of view, they form a 'natural' set or logical consequence of the new analysis. Rather, the principle of saliency, possibly along with other universal principles or language-particular restrictions, gov- erns the introduction of new elements. In graphic terms, syntactic change begins at a corner, thence spreading inward, i.e. from a region of zero differentiation between the old and the new constructions to regions of successively greater surface differentiation. In this way, the new structure and its associated properties can be partially blocked for a certain period of time. It is only in the long run, not in its initiation or progress, that linguistic change produces relatively uniform and 'natural' results.

REFERENCES BRAGA, MARIA LUIZA. 1977. A concordancia de nimero no sintagma nominal no Triangulo

Mineiro. PUC/RJ master's dissertation. CAMARA, JOAQUIM MATTOSO. 1965. A propos d'un vulgarisme du portugais du Bresil.

Omagiu lui Alexandru Rosetti, 534-35. Bucharest: Editura Academiei Republicii Socialiste Romania. [Portuguese translation in Dispersos de J. Mattoso Camara Jr., ed. by Carlos Eduardo Falcao Uch6a, 89-93. Rio de Janeiro: Editora da Fundagao Getfilio Vargas, 1975.]

--. 1972. The Portuguese language. Translated by Anthony J. Naro. Chicago: Uni- versity of Chicago Press.

CEDERGREN, HENRIETTA J., and DAVID SANKOFF. 1974. Variable rules: Performance as a statistical reflection of competence. Lg. 50.333-55.

97




GuY, GREGORY R. 1975. Variation in the group and the individual: The case of final stop deletion. (Pennsylvania working papers on linguistic change and variation, 1:4.) Philadelphia: US Regional Survey.

KAY, PAUL. 1978. Variable rules, community grammar, and linguistic change. Linguistic variation: Models and methods, ed. by David Sankoff, 71-83. New York: Academic Press.

--, and CHAD K. MCDANIEL. 1979. On the logic of variable rules. Language and Society 8.151-88.

KIMBALL, JOHN. 1973. Seven principles of surface structure parsing in natural language. Cognition 2.15-48.

LABOV, WILLIAM. 1969. Contraction, deletion, and inherent variability of the English copula. Lg. 45.715-62. [Revised version in his Language in the inner city: Studies in the black English vernacular, 65-129. Philadelphia: University of Pennsylvania Press, 1972.]

LEMLE, MIRIAM, and ANTHONY J. NARO. 1977. Competencias basicas do portugues. Rio de Janeiro: Fundacao Movimento Brasileiro de Alfabetizacao. [Available from MOBRAL-CETEP/CEDOC, Ladeira do Ascurra 115, Cosme Velho, Rio de Ja- neiro, Brasil.]

NARO, ANTHONY J. 1976. The genesis of the reflexive impersonal in Portuguese: A study in syntactic change as a surface phenomenon. Lg. 52.779-810.

--. 1978. The SWAMINC4 program. Unpublished, Departamento de Letras, PUC/ RJ.

--. 1980. Review article on Linguistic variation: Models and methods, ed. by David Sankoff. Lg. 56.158-70.

-- , and MIRIAM LEMLE. 1976. Syntactic diffusion. Papers from the Parasession on Diachronic Syntax, 221-39. Chicago: CLS. [Reprinted in Ciencia e Cultura 29.259-68.]

NIE, NORMAN H., et al. 1975. SPSS: Statistical package for the social sciences. 2nd ed. New York: McGraw-Hill.

ROUSSEAU, PASCALE, and DAVID SANKOFF. 1978. Advances in variable rule methodology. Linguistic variation: Models and methods, ed. by David Sankoff, 57-69. New York: Academic Press.

SANKOFF, DAVID. 1975. VARBRUL2. Mimeographed, Universite de Montreal. SCHERRE, MARIA MARTA PEREIRA. 1978. A regra de concordancia de nfimero no sintagma

nominal em portugues. PUC/RJ master's dissertation. VASCONCELLOS, J. LEITE DE. 1901. Esquisse d'une dialectologie portugaise. Paris: Ail-

laud. [2nd ed., ed. by Maria Adelaide Valle Cintra. Lisboa: Centro de Estudos Filologicos, 1970.]

VOTRE, SEBASTIAO JOSUE. 1978. Variacao fonologica na fala dos mobralenses da area do Rio de Janeiro. PUC/RJ doctoral dissertation.

-- . 1979. Phonological and syntactic aspects of denasalization in spoken Brazilian Portuguese. Paper presented at NWAVE8, Universite du Quebec a Montreal.

[Received 20 March 1980.]

98



[Artigo] the Social and Structural Dimensions of Syntatic Change - Naro

Documents

syntactic change anthony

model of syntactic change

syntactic changeauthors

structural dimensions

jstor terms

jstor archive

terms conditions of

structural portion