[Editor's note: Due to the font limitations of HTML, j will be used to represent a j with a wedge (hacek), and a lowered ₃ will be used to represent a yogh.]

As is well known, the first names of UCSC linguistics faculty exhibit a striking tendency to begin with the segment [j]. This paper offers an analysis that not only accounts for the local phenomenon but has far-reaching implications for the field as a whole.

(1)	Judith		Daniel
	Jorge		Sandra
	Junko		Donka
	James		William
	Jaye		Armin
	Jennifer
	Geoff1

As the data in (1) show, 6 of the 12 full-time faculty members have first names beginning in [j]. This is far too high a proportion for the situation to have arisen by chance.² A historical explanation also appears unlikely; the available records do not mention any consideration of the phonological characteristics of applicants' first names in the hiring process (with the possible exception of Jaye "[je]" Padgett). It seems that we must look deeper for an explanation.

Our account uses a streamlined version of Optimality Theory, omitting for ease of exposition the ranking arguments, specification of candidates, and tableaux. We introduce a few self-explanatory technical innovations. As a first attempt at accounting for the overall [j]-initial pattern, we propose the following markedness constraint, similar in form and spirit to HAVEONSET:

(2)	HAVE [j]-INITIAL FIRST NAME IF YOU'RE A LINGUIST

What other constraints come into play? In looking at the data in (1), an obvious first generalization emerges. Fully 60% of the non-[j] names belong to native Europeans. Let us suppose then that the following context-sensitive markedness constraint outranks HAVE -[j]:

(3)	*[j]-INITIAL NAMES FOR EUROPEAN-BORN LINGUISTS

Under this view, Daniel, Donka, and Armin could be underlyingly Janiel, Jonka, and Jarmin, with [j] going to [d] (or Ø) under pressure from (3). (We omit tedious detail here; it should be obvious to the competent reader that the shape of the underlying forms as well as the simplification to [d] or Ø follow from very general principles.) However, there is a problem under this approach with the cases of Geoff and Jim, which should be subject to the constraint in (3), giving Deoff and Dim. Since these outputs are not observed, we must reconsider the constraints.

A slightly narrower reformulation of (2) avoids the problems:

(4)	HAVE [j]-INITIAL NAME IF YOU'RE A LINGUIST AND NATIVE ENGLISH SPEAKER

Now (3) can be dispensed with entirely.

The analysis thus far has, with remarkable economy, succeeded in accounting for the regularity in the names of most of the native English speakers (Judith, Jorge, James, Jaye, Geoff). The constraint does not apply to Daniel, Donka, Armin, and Junko, which are assumed to be subject to other constraints that will not concern us here.³

The cases that have not been accounted for so far are Sandra and William (a.k.a. Sandy and Bill). We hypothesize that these cases fall together in a class we will characterize as [+oddball]. This class is subject to a constraint along the lines given in (5):

(5)	*CHANGING INITIAL SEGMENT TO [j], EVEN IF A HIGHER-RANKED CONSTRAINT SAYS YOU SHOULD, IF YOU'RE [+oddball]

The [+oddball] characterization is admittedly somewhat stipulative, but we are optimistic that further research will provide additional motivation.⁴ In the meantime, there is independent evidence from syntax showing that members of the [+oddball] class do pattern together and are in fact odd. Though in the usual case linguists routinely undergo movement from Cowell to Stevenson, linguists bearing [+oddball] names steadfastly resist such movement. We can think of this as a kind of island sensitivity. Whatever its source, it is a useful diagnostic for the [+oddball] category.

We have thus accounted for the distribution of [j]-initial names in an intuitively satisfying way. But this account, while providing an elegant technical solution, raises a deeper question. Why should we expect to encounter such name-oriented constraints in the first place?

We speculate that the answer to this question involves learnability. Note that faculty names are among the first elements acquired by graduate students. Astonishingly, graduate students typically control faculty first names even before beginning to construct their first tableaux. The speed and facility with which this feat is performed make it highly improbable that sheer memorization is involved. We hypothesize that the innate principles governing the phonological make-up of faculty names are behind this seemingly effortless mastery.

In conclusion, let us note that we have stated our proposal in maximally general terms. According to (4)–(5), we make the strong empirical prediction that all linguists who are native speakers of English either have first names starting with [j] or are inclined to resist Cowell-Stevenson movement. How does this prediction square with the facts? A quick look at the LSA membership list is encouraging. While there are quite a few first names beginning with segments other than [j], it is likely that some of these do not belong to native English speakers. Besides, some dues-paying members may not actually be professional linguists. More importantly, the possibility that the apparent "counterexamples" are actually [+oddball] cannot be ruled out. This is an empirical question, of course, even though conducting the Cowell-Stevenson island test on a large scale is subject to practical difficulties. Many of the difficulties could be addressed with an adequate research budget. However, even if the funds were available, the incompleteness of our understanding argues against attempting a premature empirical evaluation. One of the obstacles is that [+oddball] linguists from other universities may very well resist transplantation to Cowell in the first place, which would be consistent in a general way with the hypothesis of island sensitivity but which would unfortunately bar straightforward application of the Cowell-Stevenson test. This difficulty should not cause us to abandon the prediction. As long as it remains testable in principle we are on solid ground.

Perhaps the larger lesson here is that myopic dependence on the data is a sure way to miss significant generalizations. After all, we can't know what the facts mean until we know what they are supposed to mean. The latter is the task of our linguistic theory.

Endnotes

¹ We will have nothing to say here about spelling idiosyncrasies (though see note 4). [Back]

² The claim could in principle be empirically verified by corpus work and statistical analysis, relying on actual frequency counts for names in the general population. We omit this step for expedience and because in this discussion, facts will not be of central importance. [Back]

³ The Donka/Daniel pattern is intriguing. As an anonymous idiosyncratic reviewer points out, the [d]-initial pattern might be considered a generalization of the [j]-initial one if we were willing to employ the standard IPA symbol [d₃]. At a suitable level of transcriptional representation (call it Write-Out), both [d₃]- and [d]-initial names could be viewed as satisfying a [d]-initial name constraint. This idea seems quite promising, but we cannot pursue it here. [Back]

⁴ We have little to say at this point about the nature of the oddness involved. It is well known that most, if not all, linguists are relatively odd, in the ordinary sense of the term. Thus, the notion of "oddness" we are invoking here must be viewed as more technical in nature. It is clear that, at a minimum, [+oddball] must be distinguished from [+idiosyncratic]. [Back]