7. Neogrammarian Sound Change.pdf

7. Neogrammarian Sound Change : The Handbook of Historical Linguistics : Blackwell Reference Online

12/11/2007 03:33 PM

7. Neogrammarian Sound Change

MARK HALE

Subject

Linguistics » Historical Linguistics

Key-Topics

grammar

DOI:

10.1111/b.9781405127479.2004.00009.x

The essential claim of the Neogrammarians regarding “sound change” was simply that it was systematic

(“konsequent”). Hermann Paul, in explicating what this label meant, states the following (Paul 1880: 69,

given here in translation):

When we speak of systematic effect of sound laws we can only mean that given the same sound

change within the same dialect every individual case in which the same phonetic conditions are

present will be handled the same. Therefore either wherever earlier the same sound stood, also

in the later stages the same sound is found or, where a split into different sounds has taken

place, then a specific cause –- a cause of a purely phonetic nature like the effects of

surrounding sounds, accent, syllabic position, etc. –- should be provided to account for why in

the one case this sound, in the other that one has come into being. 1

This passage contains the two assertions which are assumed in modern literature on change to be the

hallmark of Neogrammarianism: sound change 2 is regular and purely phonetically conditioned . As the

insightful work of Hoenigswald (1978) has shown, these claims are true, within the context of the work of

the Neogrammarians themselves, by definition. Ancillary hypotheses (e.g., “ analogy,” the phrase “within the

same dialect,” and the concept of “sporadic sound change” 3 ) allow the Neogrammarians to restrict the use of

the term “sound change” to precisely those events which are regular and phonetically conditioned.

The question naturally arises, for the modern linguist armed with a vastly different conception of the nature

of the object of linguistic study (“the grammar”), of whether or not the terminological distinction made by

the Neogrammarians can be given a more substantive foundation –- that is, whether the distinction between

“sound change” and “sporadic sound change” can be made to follow from our current conception of the

human linguistic endowment and the nature of language change. In this chapter, I will argue that the

Neogrammarians were correct in distinguishing between two fundamentally different types of event which

can occur in the transmission of human language. Whether or not regular sound change is “purely

phonetically conditioned” also turns out to involve a number of definitional matters which will also be

addressed in what follows.

1 Diachronic Modularity and “Change”

Perhaps controversially, I will adopt a view of “change” which I believe allows one to keep distinct the various

factors which give rise to the historical record of a given language. 4 I will distinguish between change proper

and the diffusion of that change. 5 While the sociolinguistic diffusion of a change is generally necessary if

that change is to become part of the historical record of a given language, it seems clear that our

http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9781405127479_chunk_g97814051274799

Page 1 of 18

7. Neogrammarian Sound Change : The Handbook of Historical Linguistics : Blackwell Reference Online

12/11/2007 03:33 PM

that change is to become part of the historical record of a given language, it seems clear that our

responsibility to account for change cannot be coherently taken as being limited to those changes which in

fact happen to diffuse. The factors which give rise to a different representation or rule in the phonological

system of an acquirer –- given by a properly constructed learning algorithm –- are not, in my view, the same

as those factors which regulate the diffusion of that new representation or rule within a community of

speakers (given, presumably, by a properly constructed sociolinguistic theory). Questions arising from the

Neogrammarian hypothesis –- particularly the “regularity” and “conditioning” issues –- are about possible

changes, not about any individual existing change. It is impossible, a priori , to ascertain whether a given

possible innovation will diffuse, since, being merely a potential event, it has no particular sociolinguistic

context. 6 To put the case more strongly, I believe that a full evaluation of the Neogrammarian hypothesis

allows one to place each potential change in any imaginable sociolinguistic context, including those from

which diffusion is virtually inevitable. 7

The historical record of an actual linguistic tradition involves, therefore, several filtering subsystems

(“modules”), as reflected in figure 7.1 . I believe that it is critical to keep these various subsystems and their

internal dynamics distinct from one another, if progress is to be made in understanding the contribution

made by each to the overall linguistic record. 8

Developing a coherent model of the first box is the central responsibility of historical linguistics. The

“diffusion” box represents the primary domain for sociolinguistic theorizing. The final box includes, among

other issues, such matters as who had access to writing (and who did not), what linguistic features the

writing system encodes (perhaps indirectly), and how these features can be extracted from the extant

record, what survives, etc. Much of this is the central subject matter of the field of “philology.” A

comprehensive understanding of the actual historical record thus requires contributions from all of these

fields (constrained by relevant synchronic theories of the linguistic modules involved –- phonological,

morphological, syntactic, etc.). The Neogrammarian hypothesis is, in my view, a claim about the first box –

the set of possible change events –- and only the first box.

Figure 7.1 Modularity and the historical record

http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9781405127479_chunk_g97814051274799

Page 2 of 18

7. Neogrammarian Sound Change : The Handbook of Historical Linguistics : Blackwell Reference Online

12/11/2007 03:33 PM

Figure 7.2 The nature of change

It is critically important to provide a precise characterization of what precisely a “change event” is. Consistent

with the “modular” approach discussed above, we believe that “change” is to be conceived of as the set of

differences between the grammar generating the primary linguistic data (PLD) used by an acquirer and the

grammar ultimately constructed by that acquirer. 9 Idealizing away from the problem of “multiple sources” we

can sketch the relevant scenario as in figure 7.2 (for discussion of earlier, similar diagrams, see Janda

2001). 10

In this figure O 1 represents the acquirer's source –- the PLD which are themselves the output of an existing

grammar G 1 ; S o is the initial state of the acquirer's knowledge (in L1 acquisition – and arguably in L2

acquisition as well –- this is Universal Grammar (UG)). The intermediate, transitory stages of the acquisition

process are represented by S1 , S 2 , etc. The end-point of the acquisition process, for this particular

grammar, is represented by G 2 : evidence which the acquirer receives after this point which is not consistent

with G 2 will not be used to “modify” this knowledge state (though it may give rise to a new acquisition

sequence, of course). “Change,” in the sense we will be using it in this chapter, is simply the differences

between G 1 and G 2 . Note that since G 2 comes into existence at some well-defined point in the acquisition

process, all “change” under this model will necessarily be abrupt. 11

2 Regularity and Phonetic Conditioning

In general, modern literature on the Neogrammarian doctrine assumes that its two central propositions –-

that sound change is regular and that it is purely phonetically conditioned –- are independent. The

propositions are thus usually evaluated in a manner consistent with that assumption. It has been claimed by

numerous modern authors that both propositions are false (see, e.g., Kiparsky 1995b, reprinted this volume

(but hereafter simply “1995”), with literature).

It is not without interest to attempt to understand why the two proposals are linked by the Neogrammarians

themselves. I believe that the Neogrammarian hypothesis represents not two independent conjoined claims

about the nature of sound change, but rather two necessarily related components of a single conception of

the phenomenon. I will demonstrate this first by showing that a standard interpretation of the meaning of

“regular” appears to be, on its own, relatively uninteresting. However, when put together with the issues

surrounding the proper characterization of the environment in which a change takes place – that is, its

conditioning – the issues become much more intriguing. While the Neogrammarians cannot, given the state

of their understanding of the nature of grammars, have had precisely the system I propose in mind, it seems

http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9781405127479_chunk_g97814051274799

Page 3 of 18

7. Neogrammarian Sound Change : The Handbook of Historical Linguistics : Blackwell Reference Online

12/11/2007 03:33 PM

of their understanding of the nature of grammars, have had precisely the system I propose in mind, it seems

that the fundamental success of methods which are directly dependent upon Neogrammarian notions, such

as the comparative method, indicate that their pre-theoretical phenomenological insight in this domain was

quite advanced. 12

Turning first to the “regularity” issue, it would appear that the standard interpretation of this term in

historical linguistics is relatively straightforward. Given a change of the type X → Y/Z, the change is regular

iff for every X in environment Z in G 1 , we find Y in G 2 . 13 It seems clear that “sound change” (and indeed, any

change) will be “regular” under this conception of what precisely “regularity” is. A change will be a maximally

general statement of a difference between G 1 and G 2 of figure 2 . If a claimed change (X → Y/Z) is a true

assertion about the relationship between G 1 and G 2 , then the conditions for the application of the term

“regular” will be met. While this appears at first to make “regularity” a resoundingly uninteresting issue, I will

attempt to show in what follows that it allows one to focus the discussion on precisely those issues most

relevant to the understanding of the “regularity of sound change.”

Let us examine a typical case of what has traditionally been called “sporadic” sound change (generally felt to

be non-regular –- i.e., outside the domain of Neogrammarian “sound laws”): Proto-Polynesian (PPN) * lango

shows up as ngaro in Maori. 14 The expected outcome, given the regular change of PPN l to Maori r, is rango

–- the attested form shows an irregular metathesis. 15 This “change” took place on at least one occasion in

the speech of someone from whom, for sociolinguistic reasons, it diffused. 16 An accurate statement of that

change at the moment of innovation will require that the environment, Z, be lexical , rather than

phonological –- that is, this was a change in the phonological representation of an individual lexical item,

not in the phonological system of Maori. Since, if the statement of the change is to be accurate, the

environment must fully spell out the lexical item in which the change took place, the change will be regular

within its domain (in this case, a single lexical item).

We see, therefore, that the environment is crucially involved in any discussion of “regularity.” If the term

“regular sound change” is to have any useful meaning, we cannot use it to refer to any change which is

regular in its stated environment (for, if the environment is stated correctly, this will always be the case). On

the other hand, we cannot require of a change that it have no conditioning environment if it is to be counted

as “regular” –- this would exclude many cases which are clearly regular in the required sense (e.g.,

intervocalic lenition, final consonant loss, etc.). One coherent way to limit the term “sound change” is thus

by requiring that the environment in which the change takes place be specified in phonological rather than

lexical terms. This was, in some ways, the tack taken by the Neogrammarians, and it seems a useful one.

Neogrammarian theory was thus never intended to account for changes in the phonological representations

associated with individual lexical items. Such “lexical” changes are rather numerous –- for example, my

grandmother's word for what I call a ‘ couch ’ was ‘ davenport. ’ This is not a change anyone would want to call

a “sound change,” clearly, even though the phonological representation associated with a given semantic

entity has changed. If we restrict “sound change,” as we, in my view, must if we are to exclude dævnport >

kawč , to instances in which the environment is to be stated in phonological, rather than (e.g.) lexical terms,

it is clear that sound change will be regular in the required sense. 17

It thus appears that fundamentally distinct types of misanalysis are involved in the two cases. It would not

be helpful to the enterprise of historical linguistics if this difference were to be ignored. To call the contrast

between “(regular) sound change” and lexical changes of the type we have discussed above “merely

definitional” entails that there is no crucial distinction in the underlying dynamic which gives rise to the two

types of change event. To the extent that there is a fundamental difference in these mechanisms the

“regularity of sound change” ceases to be a purely terminological matter.

3 The Causes of Change

Given the notion of “change” we have adopted, the possible causes of change are highly restricted. 18 A more

detailed examination of the acquisition process will reveal why this is so. The components involved are

illustrated (in a schematic way) in figure 7.3 .

In figure 7.3 , A represents the interpretation of the output of the grammar (a mental representation) by the

articulatory performance system (including all relevant cognitive and physiological systems). 19 In keeping

http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9781405127479_chunk_g97814051274799

Page 4 of 18

7. Neogrammarian Sound Change : The Handbook of Historical Linguistics : Blackwell Reference Online

12/11/2007 03:33 PM

articulatory performance system (including all relevant cognitive and physiological systems). 19 In keeping

with widespread assumptions, I take this mapping to be universal, though perhaps “chaotic” in the technical

sense. 20 B represents the various transformations that the actual acoustic signal which results from a given

articulatory act will undergo. These are generally contingent upon environmental factors and design

properties of the organism performing the articulatory act (size of resonating cavity, force and direction of

ambient airflow, etc.). Though no doubt universal, these factors are clearly “chaotic.” An essentially random

subset of output of B (that portion the acquirer is in a position to hear) must then be processed by the

speaker's perceptual system –- this transformation is indicated by C in figure 7.3 . Again, I will assume this

mapping to be universal and, arguably, chaotic. The portion of the resulting perceptual process deemed by

the learning algorithm (D) to be relevant to the acquisition task will be treated as the PLD by the acquirer.

The learning algorithm itself, uncontroversially, will be taken as deterministic and universal.

Processes internal to the source grammar (e.g., phonological rules and representations) are not directly

accessible to the learner. One of the key questions confronting a theory of language acquisition is just what

types of inference the acquirer can draw from the quite indirect evidence for these processes that is present

in the PLD. Only those aspects of these processes which are reflected in some manner in the PLD can be

acquired. If the mapping from the PLD to the grammar is deterministic, as I am assuming, then aspects of

the source grammar unambiguously reflected in the PLD will normally imply acquisition of those features

(i.e., no change along the relevant dimension).

Figure 7.3 Sources for “noise” in the PLD

These considerations lead to the following conclusion: change can only result from the acquirer being

exposed to primary linguistic data (PLD) which differs in some way from the PLD which were presented to

the source during the source's own acquisition. There are two primary forces giving rise to such differences

(see Hale 1998, 1997 and Ohala 1981a for extensive discussion):

i The unique subset of data presented to the acquirer of G 2 may, and in virtually every case will, be

different from that presented to the acquirer of G 1 either in scope or in sequence , or both (cf. Janda

1990, 1994a).

ii The acquirer may mistake the effects of the speaker's production system (A), of ambient effects on

the acoustic stream (B), or of his or her own perceptual system (C) as representative of G1-internal

representations or computations. These are the “noise”-introducing factors sketched in figure 7.3 .

The factors shaping the PLD given in (ii) above are contingent and sporadic in their effects (if systematic

factors of these types exist, it appears that they can be filtered out as irrelevant by the acquirer). While

chance distortion introduced by any of these factors may impact the acquirer's ultimate representation of a

given lexical item (and thus lead to “lexical” change of the type discussed above), they are too context-

dependent to give rise to regular phonological change. Only the factors in (i) –- the finiteness and order of

presentation problems –- should be relevant to what we have called “regular phonological change.”

The “finiteness” problem is potentially relevant in the following sense: while the total output of the source

grammar with respect to some phonological sequence may provide more than sufficient evidence for an

unambiguous parse of that sequence, the acquirer gets only a subset of the evidence. The idea here is

relatively simple: the realization of phonological targets by the source grammar speaker will look like a

scatter diagram, with a mid-point in that position in the acoustic space which most unambiguously reflects

the true nature of the target involved. For example, in realizing an aspirated voiceless labial stop, specific

instantiations will have different temporal durations for the stop and for the aspiration, or the stop may

http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9781405127479_chunk_g97814051274799

Page 5 of 18

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika: