Me-muh-ry or mem-ry (memory), fa-muh-ly or fam-ly (family), ce-luh-ry or cel-ry (celery), and so on – there are different ways to pronounce words. Across languages, linguists spend a lot of time figuring out what it is about the way that we use language that predicts when speakers produce certain words.


UNM Ph.D. Linguistics student and Joseph Greenberg Fellow Sarah Lease has just published research that contributes to answering this question.  

“I think this study for me was a great way to kind of dip my toes into working on this question and simultaneously being able to provide to the field of New Mexican Spanish information on a new variable,” Lease said.  

Lease has just released the article A usage-based account of paragogic /e/ in 20th century New Mexican Spanish in Studies in Hispanic and Lusophone Linguistics. She aimed to pinpoint variables that predict when New Mexican Spanish speakers produce a hallmark feature of this dialect and when they don’t.    

Similar to the fam-ly versus fa-muh-ly alternation, Lease focused on alternations in words that display paragogic /e/ - the addition of the vowel /e/ at the end of a word. In New Mexican Spanish, this alternation is found in words like rincón and rincone (‘corner’) and leer and leere (‘read.’) 

“These are the same exact words for speakers–it’s just two ways of producing these words,” Lease said.  “The paragogic /e/ just hasn't received a lot of attention in previous scholarly work, so I was just kind of interested in trying to expand the body of work on New Mexican Spanish by looking at this less studied feature.”

The study focuses on the variable of paragogic /e/ (rincón or rincone ‘corner’) among New Mexican Spanish speakers, which provides an exceptional test case for the consequences of words’ cumulative usage in paragoge’s conditioning contexts.

This follows along with the efforts of other linguists at UNM who strive to preserve and document New Mexican Spanish.  

For the project, Lease searched for words that alternated between a paragogic /e/ rincone and non-paragogic-/e/ form rincón in UNM data of New Mexican Spanish that had been compiled by other UNM researchers. These corpora included the New Mexico and Colorado Spanish Survey, the New Mexico Spanish-English Bilingual corpus, and Las Pláticas.   

“It was a combination of reading all of the transcripts, identifying cases of paragogic /e/ and then going back and listening to those recordings to kind of confirm its existence,” Lease said.  

Lease found 77 types of words that variably displayed paragogic /e/ produced by 24 speakers. She looked at 2,235 instances of these words. 

“First, there were some social factors that predicted when speakers produced paragogic /e/ and when they do not. For instance, it depended on where they were in New Mexico,” she said.

In that data she found 76% of speakers who hailed from Rio Arriba County produced the paragogic /e/. That was not the case when looking at San Juan County or Santa Fe County. 

Also, as forewarned by previous UNM research, younger generations were not as likely to maintain traditional New Mexican speech. In Lease’s analysis, all paragogic /e/ users in the sample were born in or earlier than 1950. 

Aside from social factors, she found that the likelihood that speakers produce paragogic /e/ depended on how often the word is used. Lease found that as words were used less and less in discourse, this paragogic /e/ variant became more likely to be produced.

Second, in the words that variably display paragogic /e/, Lease found that speakers most often produce paragogic /e/ at the end of a thought.  

“There is a significant effect of the following context on this particular alternation, such that when these words are produced at the end of a phrase, they're more likely to be produced with paragogic /e/,” Lease said.  

That means that for a word like agrimensor or agrimensore (‘surveyor’), New Mexican Spanish speakers would be more likely say yo trabajo con un agrimensore (‘I work with a surveyor’) rather than yo trabajo con un agrimensor (‘I work with a surveyor’.) While speakers could say something like el agrimensore llegó ayer (‘the surveyor arrived yesterday’), it's much less likely since agrimensor isn’t at the end of the phrase.   

“I also used a nuanced measure of language use to demonstrate how speakers’ accumulated experience with language affects variation in these words. This cumulative experience influences the likelihood that speakers produce paragogic /e/,” Lease said. 

The metric, referred to as FCC in her study, estimated the frequency with which words are produced at the end of a phrase, which is the discourse context where speakers most likely produce the paragogic /e/.  

This is a common tendency in language. Speakers were more likely to produce the paragogic /e/ in words used less often, like sale or sal (‘salt’), than in more frequently used words, like españole or español (‘Spanish.)

With lower frequency words, this variant, this paragogic /e/, form is more common. It also happens to be the case that the more often these words occur in the phrase-final position, the more likely they are overall to be produced with the paragogic /e/ form.

“I think that's an important implication and shows that we really know so little about these FCC effects. And we just need to keep using them to really understand where they operate,” she said.

Owing to speakers’ cumulative experience with paragogic /e/ words in phrase-final position, the likelihood of producing paragogic /e/ also increases in other discourse contexts where paragogic /e/ typically isn’t produced. This type of effect explains why speakers sometimes do produce el agrimensore llegó ayer (‘the surveyor arrived yesterday’), where the paragogic /e/ is not in the discourse context expected.

“This whole idea of the effect of cumulative usage on phonetic variation is very interesting, and it goes a long way in explaining the words' specific ranges of variation,” Lease said.

Lease is encouraging current and future linguists to use more nuanced measures of language use, like the FCC metric. 

“We still don't know a lot about it yet and we don't know all of the implications that it has in terms of speech production,” she said. “The more times people apply this metric, the more we can learn about what it says about variation in speech production.”