Printer Friendly


1. INTRODUCTION. There exists a tendency in sociolinguistic analyses to confuse or combine FAST SPEECH and CASUAL SPEECH. Hasegawa (2006:19) notes that 'casualness and the rate of speech are often interchangeably used to refer to certain phonological processes that take place in [a] RELAXED or CASUAL speech context' (emphasis in original). This tendency has not been widely questioned in the past, as the two speech types overlap a great deal. However, the general acceptability of this practice does not necessarily make it the correct practice.

It is true that casual speech is often fast and that fast speech is often casual, but these generalizations are not categorically accurate. As attested by Zwicky (1972:607), 'some speakers .. use a quite informal style even at fairly slow rates of speech, while others... give the impression of great precision even in hurried speech.' In fact, beyond individual propensities that call into question the generalization, the difference between fast speech and casual speech processes is even more apparent in certain languages with greater sociological sensitivities. With 'sociological sensitivities,' Hasegawa (2006) refers to the fact that the Japanese language is highly attuned to vulgarity, formality, familiarity, and intimacy. According to Hasegawa (1979, 2006), grammatical processes in Japanese are inextricably tied to these social factors in ways that in other languages they are often not, making the distinction between fast and casual speech processes in Japanese much more salient. Hasegawa concludes that the principle difference between fast and casual speech processes is whether a process is susceptible to rate of speech: fast speech processes are, while casual speech processes are not.

The present work seeks to determine to what extent casual speech differs from fast speech in languages like Spanish that are less attuned to sociological sensitivities (outlined in Section 2.3). More specifically, I utilize recent data from Elche Spanish (Chappell & Martinez Ibarra 2017), a variety of Valencian Spanish, to establish more concretely if the tendency to conflate fast and casual speech in linguistic analyses is warranted and methodologically sound. This study draws upon quantitative findings rather than relying upon introspection, which is an important contribution to the conversation, as the difference between fast speech and casual speech may be overlooked precisely because of its opacity to introspection. Ultimately, this quantitative analysis results in an updated approach that helps predict variable phonetic reductions based on fast speech, on the one hand, and casual speech, on the other.

The organization of this article is as follows. Section 2 provides a review of the literature on fast and casual speech, including the definitions and general conflation of the two processes in previous work (2.1) and the distinctions made by Hasegawa (1979, 2006) between fast and casual speech processes in Japanese (2.2). Section 2.3 discusses the sociological sensitivity of Spanish and Catalan compared to Japanese, and 2.4 reviews /s/ rhotacism in Elche, Spain. Sections 3 and 4 detail the study's methodology and the behavior of rhotacism in Elche, respectively. A revised approach is proposed in Section 5, and Section 6 summarizes the importance of this work and provides concluding remarks.


2.1. DEFINITION AND TREATMENT OF THE TERMS. Fast speech is easier to delimit than casual speech, as it is directly related to speech rate. That is, speech becomes faster as the number of syllables produced per second increases (Kendall 2009, Garcia 2015). Although this definition seems straightforward, references to 'fast speech' are often subjective, as this may be used as an offhand descriptor without objectively measuring differences in speech rate. Casual speech, on the other hand, is a more elusive concept, and it is often defined in relation to context, formality, and attention paid to speech. According to Labov, casual speech is considered 'the normal speech used in informal situations, where no attention is directed to language' (1964:167). Therefore, fast speech is technically independent of context and formality, and casual speech is technically independent of velocity.

Conflations of fast and casual speech abound in the literature, ranging from older publications based on introspection to recent theoretical and quantitative works. Zwicky (1972:607) explains that 'in general, casual speech is fast', and Browman and Goldstein (1990:16) specify that casual speech is a 'subset of fast speech in which reductions typically occur.' Such phonological reductions in fast and/or casual speech are usually described as the result of articulatory undershoot (Lindblom 1963), as 'the articulators simply cannot achieve their targets in the time available' (Zwicky 1972:608). Crucially, it is precisely 'in faster, casual speech, [that] we expect gestures to show decreased magnitudes (in both space and time) and to show increasing temporal overlap' (Browman & Goldstein 1990:17, emphasis mine). The use of fast and casual either synonymously or hyponymously in a range of publications is telling; the differentiation appears to be inconsequential to some scholars. As a result, 'the delineation of registers or tempos continues to be impressionistic' (Zwicky 1972: 607), which problematizes sociolinguists' attempts to clearly and concisely define the factors predictive of a reduced variant.

It seems the delineation between casual and fast speech has been historically murky in part because of the difficulty defining each category. Another potentially confounding factor is Labov's Observer's Paradox (1972:209), which notes that we cannot observe something without changing it. In the case of linguistic observations, the act of observing casual speech makes it inherently less casual. Although certain tools may be implemented to decrease the attention an individual pays to her/his speech, it is difficult to know (i) just how casual we can consider a speech sample on the continuum of casual to formal speech and (ii) to what other samples of casual speech we can reasonably compare said speech sample.

To complicate matters even further, there are several cases in which rate of speech itself is difficult to define, as it overlaps with other important categories in sociolin-guistic analyses. If we turn our attention to investigations of word-final consonant reduction, we find that several studies have taken into consideration the number of syllables in the target word as a factor predictive of final consonant reduction (Chappell 2015a, Dohotaru 1998, Ruiz-Sanchez 2004, Terrell 1979), as reduction is more likely in words with a greater number of syllables. However, the disentanglement of syllable number and speech rate is quite a challenge, as syllables are generally shorter in duration in longer words, with drastic reductions taking place in words with a higher number of syllables (Johnson 2004). In a case like this one, in which factors like syllable number and speech rate are so grossly intertwined, how are we to clearly define or determine the effect of one of these factors independently?

Given the difficulty in (i) determining what, exactly, constitutes casual speech and (ii) separating fast speech from other factors potentially predictive of lenition, it does not come as a surprise that the distinction between casual and fast speech has been blurred in previous discussions. As File-Muriel and Brown (2011:224) explain, 'the influence of register can often be difficult to disentangle from speaking rate, given that informal registers are often characterized by more accelerated speaking rates than are found in more formal registers.' The interconnectedness of these categories has prompted rather muddled analyses pointing to register, speed and/or formality as a singular explanation for higher rates of lenition. For example, in his investigation of /s/ lenition in Central American Spanish, Lipski (1985) claims that register, style, and speech rate may all be factors in [h] or O production. In many linguistic analyses, the line between casual and fast speech may be opaque, unspecified, or considered unimportant.

The conflation of two distinct scales into a single generalized category due to their tendency to overlap harkens back to Milroy 's (2001) discussion of linguists' tendency to view 'prestige' and 'standard' variants as one and the same, utilizing either label without distinguishing between the two. As Milroy adeptly explains, standardization is the imposition of a norm over a class of objects, e.g., the standardization of currency or weights, while prestige is defined by overtly positive social evaluations toward a variable or variety. With languages like English, Spanish, French, Catalan, etc., the most 'standardized' variants and varieties tend to receive the highest social evaluation, or prestige, but this is not to say that prestige forms an inherent part of the standardized variety, nor would it motivate the concepts' conflation. For example, Carvalho (2006) explains that the standard pronunciation of coda /s/, e.g. [bas. ta] 'enough,' is stigmatized in Rivera, Uruguay due to its negative association with border Spanish influenced by Portuguese, while the nonstandard [h], e.g. [bah.ta], is considered the prestige variant due to its associations with the 'uncorrupted' monolingual Spanish spoken in Montevideo. Although standard and prestige variants are not necessarily the same, they tend to be conflated due to their frequent co-occurrence. In a similar fashion, casual speech and fast speech tend to co-occur and, as a result, are often conflated in spite of their differences.

To be clear, this is not to say that there is no relationship between the prestige and the standard, on the one hand, or fast speech and casual speech, on the other. Previous work has found quantitative support for the overlap between fast and casual speech, with slower speech rates in reading tasks than in spontaneous, more casual speech (Jacewicz & Fox 2010, Tauroza & Allison 1990, among others). However, fast speech does not categorically lead to the same reduction patterns in speech, and many differences have been found in how speech rate affects gestural overlap (Adams et al. 1993, Kuehn & Moll 1976, Lubker & Gay 1982, Matthies et al. 2001, Nolan 1992, Shaiman et al. 1997, among others.) In fact, different strategies may emerge in fast speech, with some speakers maintaining the same phonological style they exhibited when producing slower speech and others exhibiting more gestural undershoot and phonological reductions in fast speech than in casual speech (Dressier & Wodak 1982:354). That is, while previous research has found a connection between fast speech and casual speech, the two are not necessarily one in the same. Hasegawa (1979, 2006) highlights this point, showing that fast speech and casual speech are unmistakably different processes in Japanese. Her conclusions are outlined in Section 2.2.

2.2. HASEGAWA'S DISTINCTIONS. TO illustrate the differences between fast and casual speech in Japanese, Hasegawa (1979, 2006) impressionistically distinguishes between phonological features that are sensitive to the rate of speech, including high vowel devoicing, and features that are sensitive to casual speech, including nasal syllabicization and vowel fusion. (1) Hasegawa argues that fast speech and casual speech condition reductions in different ways, and the distinction between them is more obvious in Japanese, which involves a deeper awareness of power and solidarity among the speaker, addressee, and referent than other languages (Park 1990). In fact, Japanese is so attuned to social and contextual appropriateness given the interlocutors' relationship that in informal contexts some grammatical processes take place regardless of speech rate. On the other hand, other processes in Japanese are more dependent on speech rate regardless of register, suggesting that the conflation of fast and casual speech is unwarranted in the Japanese language.

Certain processes in Japanese appear to be the simple result of gestural overlap in rapid speech. For instance, high vowel devoicing takes place in fast speech between voiceless consonants (or with a preceding voiceless consonant and a following pause), shown below in (1).

(1) a. /kisetw/ [right arrow] [kisetw] 'season'

b. /sikasi/ [right arrow] [sikasi] 'but' (Hasegawa 2006: 171)

Such a process can be explained in terms of articulatory phonology (Browman & Goldstein 1986, 1989, 1990), an approach that uses gestures as phonological primitives and explains coarticulation and reduction as the result of gestural coordination and overlap. As speech rate increases, gestural overlap increases, and the open glottis gesture of the adjacent sounds begins to extend to the neighboring vowel. Such gestural overlap prevents the vocal folds from reaching vibration next to or between voiceless consonants (see Jun & Beckman 1993; Beckman 1996), a configuration depicted below in slow speech (Figure 1) and fast speech (Figure 2).

While successfully explaining high vowel devoicing, timing constraints alone cannot account for all reductions in Japanese. For example, the nasal syllabicization of /no/ (2) and /ni/ (3) requires a specific phonological context, occurring only with a following voiced alveolar sound. Additionally, 'the application of [nasal syllabicization] is lexically specified as to what grammatical categories ni, no, and the items that follow them should be. That is, the no that is nominative or genitive can be syllabicized only when it is followed by the copula, and the copula ni when it is followed by the predicate nar-u "become"' (Hasegawa 2006:174-75), adapted from Hasegawa below in (2).

(2) Nasal syllabicization with /no/ and /ni/{/no/, /ni/ [right arrow] [n] /__{/n/, /d/}

where /ni/ = copula, /no/ = NM or GEN; and {/n/, /d/} = COP or nar-u

e.g. /kuru no nara/ [right arrow] [kuru n nara] (174-175).

In other words, the grammatical requirements for nasal syllabicization are far more complicated than the relatively straightforward case of high vowel devoicing. With nasal syllabicization, other linguistic specifications such as word class and following phonological segment help guide the reduction, which is conditioned by context rather than speech rate alone.

Vowel fusion is another process insensitive to rate of speech in Japanese. According to Hasegawa (2006:178), vowel fusion 'takes place under much more restricted sociological conditions. Primarily in men's vulgar or unrefined speech, (4) /ai/, /oi/, and /ae/ are fused into a long vowel [e:]' (178), illustrated below in (3).

(3) a. /nai/ [right arrow] [ne:] 'not'

b. /omae/ [right arrow] [ome:] 'you'

The two aforementioned reductions arise in both fast speech and slow speech, variably occurring whenever the context is sufficiently informal. In other words, variable reductions in Japanese are not exclusively guided by fast speech, which yields less time for articulation and prompts gestural undershoot. Rather, Hasegawa contends that some reductions are attributable to sociological factors rather than rate of speech.

Hasegawa (2006:173, emphasis in original) proposes that fast speech processes 'are phonologically NATURAL', sensitive only to phonological environments and do not 'indicate any obvious change in the sociological level or stylistic context.' We can expect, then, that '... as the speed of the speech of announcers, conference presenters, and so forth, gets accelerated under the pressure of running out of time, the number of occurrences of devoiced vowels would increase' (Hasegawa 2006:172), irrespective of the sociological context. However, casual speech processes would not follow the same pattern.

According to Hasegawa (2006:187), casual speech processes are more restricted than fast speech processes, applying only to certain lexical items within particular structures, when the context corresponds to certain sociological notions, and the casual speech process may or may not apply depending on other grammatical components in the utterance. Hasegawa's data seem to resolve the issue in Japanese; fast speech and casual speech are not one and the same, rendering their conflation inappropriate. What remains unresolved is whether this conclusion is also applicable to other languages with lesser sociological sensitivity. Would the same differences for fast speech and casual speech be found in a language like Spanish? To answer this question, we follow Hasegawa's (2006) terminology and turn to a brief discussion of these languages' 'sociological sensitivity.'

2.3. THE CONCEPT OF 'SOCIOLOGICAL SENSITIVITY'. The Japanese language reflects the sociological sensitivity of its corresponding culture through a complex honorific and humilification system (Kim & Sells 2007, Song 2008). According to Reischauer (1995:126) and other Japanologists, Japan represents a 'vertical society' in which even small differences in status are considered highly important. Communicative axes guide the appropriate use of honorifics in Japanese, meaning the relationships among speaker, addressee, and referent are taken into account, and two general concepts, power and solidarity, contribute to the selection of the appropriate honorific. Park (1990:113) states that POWER involves the 'relative superiority of one person over another,' while SOLIDARITY '... is defined as relative intimacy or distance between/among individuals.' Each concept involves two subgroupings of honorific variables, including ascribed power and achieved power along the power dimension and interpersonal solidarity and situational solidarity along the solidarity dimension. ASCRIBED POWER revolves around social variables of age, kinship, and sex, while ACHIEVED POWER accounts for social status and occupational rank. INTERPERSONAL SOLIDARITY considers individual and group variables like intimacy and membership within a group, respectively, and SITUATIONAL SOLIDARITY applies to the degree of formality or casualness (Park 1990:111-16). Neither Spanish nor Valencian Catalan encode such a detailed array of hierarchical, interpersonal, and contextual variables in the languages, and, as a result, they are considered to be less sociologically sensitive than Japanese.

Of course, this is not to say that Spanish and Valencian Catalan require no sociological sensitivity; rather, the idea is simply that these languages, and Romance languages in general, are not as highly attuned to the myriad social distinctions encoded in the Japanese language. Many features in Spanish do indicate awareness of sociological factors, the most salient of which would likely be considered the T-V distinction in pronominal forms of address. The use of tu 'you (singular, informal)' and usted 'you (singular, formal)' requires an understanding of social distance, age, occupation, and myriad other extralinguistic factors (see Sinnott 2010), and a similar informal-formal distinction exists in Catalan with the pronominal forms of address tu and voste (Wheeler et al. 1999:161).

Both Peninsular Spanish and Catalan, however, seem to have undergone changes regarding their sensitivity to power and hierarchy in the past century. In Catalan, for example, another polite pronominal form of address that involves less distance than voste, vos (cf. French vous), was systematically used before the Spanish Civil War (1936-1939), but the pronoun is now relegated to specific addressees and contexts, e.g. elderly addressees and in business correspondences (Wheeler et al. 1999:162). Similarly, the use of the formal usted form has decreased dramatically in Spain in the post-Franco era, and today the tu form is more widespread in Spain than anywhere in Latin America (e.g., Carricaburo 1997).

Stewart (1999:121) describes this development as an 'unparalleled shift towards the use of the familiar second person tu' at the expense of the usted form. This shift toward informality is even clearer for second-person plural pronouns, where the informal plural vosotros form is used more readily than the formal ustedes, even in contexts that would require the formal usted form for a single referent (Morgan & Schwenter 2016). In this sense, at least, the importance of verticality in Peninsular Spanish seems to be decreasing. In fact, Brown and Gilman (1960) argue that the POWER SEMANTICS that predominated in Spanish until the 19TH century have been supplanted by SOLIDARITY SEMANTICS, and this increased emphasis on solidarity is reflected in pronoun use.

This dampened sensitivity to differences in status and formality presents a sharp contrast to Japanese, which may yield differential treatments of linguistic variables in fast speech or casual speech. The next section explores one linguistic variable that can help shed light on this issue: coda /s/ in Elche Spanish.

2.4. LANGUAGE USE AND RHOTACISM IN ELCHE. Located along the Mediterranean Coast in southeastern Spain, Elche is the third largest city of the Autonomous Community of Valencia, with a population of approximately 230,000 residents called ilicitanos (Instituto Nacional de Estadistica 2013). Both Spanish and Valencian, a dialect of Catalan, are spoken in Elche, but research suggests that the centuries-long language shift away from vernacular Valencian and towards standard Spanish continues, even in the face of renewed linguistic planning (Gimeno-Menendez & Gomez Molina 2007:99-101). Although nearly all ilicitanos attest that they speak Spanish frequently on a daily basis (97%), only 25% claim to speak Valencian frequently on a daily basis (Martinez Ibarra 2013). Additionally, Spanish is used in a wide range of public and private contexts, while Valencian is reserved for private domains like the home or in specific local interactions.

Native Spanish speakers in Elche may rhotacize /s/ ([r] or [[phrase omitted]]) before a voiced consonant or before a vowel, e.g. los goles 'the goals' and los otros 'the others' as [lor voles] and [lor otros]. This process is generally said to take place where /s/ can be realized as [z] and can be described as a case of articulatory undershoot (Kapovic 2015, Lindblom 1963). Both [z] and [r]/[[phrase omitted]] are alveolar, and if /s/ is produced with a reduced duration or insufficient pressure to create frication, a flap or approximant rhotic may be produced (Romero & Martin 2003). In other words, as more gestural overlap takes place, the degree of constriction changes and yields a rhotic instead of a fricative.

Perceptual confusion supports the similarity between [z] and [r]/[[phrase omitted]] (Painter 2011, Sole 1992). Children tend to replace [z] with a rhotic in certain developmental stages (Smith 1973:79, 262), and American English speakers perceive a rhotic when the duration of [z] is reduced, e.g. listeners claim to have heard littered [[phrase omitted]] when the duration of the [z] in lizard [[phrase omitted]] has been shortened (Romero & Martin 2003). Finally, [z] and rhotics are acoustically similar in terms of their formant transitions, duration, and intensity (Painter 2011). The reduction of [z] to a rhotic has been attested in Germanic and Sanskrit languages (Painter 2011, Roberts 2012, Sole 1992), which again supports the phonetically motivated nature of the reduction process: gestural overlap can yield a rhotic rather than a fricative.

Although no attitudinal studies have been conducted to measure the social perceptions of /s/ rhotacism in Elche, Francisco Martinez Ibarra, a linguist and native speaker of the variety, notes that 'El fenomeno parece que esta algo estigmatizado... Los que si identifican que existe te van a decir que se asocia con gente con poca educacion formal y/o del campo.' 'The phenomenon seems to be somewhat stigmatized... Those who can identify that it exists will tell you that it is associated with people with little formal education and/or from the country' (p.c. 5/29/2014). (5) That is, Martinez Ibarra's native speaker intuitions and his interactions with other speakers from the region indicate that /s/ rhotacism, particularly rhotacism before a vowel, is a variant that receives an overtly negative social evaluation in Elche.

3. METHODOLOGY. To determine in what linguistic and social contexts the rhotacized variant emerged, Chappell and Martinez Ibarra (2017) investigated the production of coda /s/ in Elche, and the dataset used in the present study comes from this corpus. In the corpus, 30 sociolinguistic interviews were conducted in August 2014. A linguist and native speaker of Elche Spanish (Martinez Ibarra) conducted these interviews in a quiet environment convenient for the interviewees, including the speakers' homes, quiet parks, or the speakers' places of employment.

Participants were recruited from a relatively even distribution of genders, ages, and education levels through the 'friend-of-a-friend' method (Milroy 1980). These participants included 16 men and 14 women with an age range between 25 and 60 years old, and there was a three-way educational distinction among participants with a college degree, a high school diploma, and without a high school diploma. The speakers came from neighborhoods within the city limits of Elche, including the city center, Carrus, Altabix, Pont Nou, and El Pia.

An Olympus WS-400S standard digital recorder was used to record each interview, and each speaker participated in two tasks. First, a casual conversation took place regarding topics of interest to the interviewees, including family, work, hobbies, movies, etc. Second, participants read a list of 79 sentences that were designed to potentially elicit rhotics, as they provided specific contexts of /s/ before voiced consonants and vowels that could result in rhotacism. Due to the ongoing language shift toward Spanish in the community, language dominance could not be controlled for, as most of the speakers were dominant in Spanish. Of the 30 speakers, all were fluent in Elche Spanish and 11 could speak Catalan as well. Of the 11 speakers of Catalan, only three acquired Catalan as their first language. The bilingualism of some participants is not expected to influence the results because all the speakers of Valencian Catalan were also native speakers of Peninsular Spanish.

After recording the participants in the casual sociolinguistic interview and the more formal reading task, 1,207 tokens of /s/ were coded for linguistic factors potentially predictive of rhotacism, including target and following word class, preceding and following segment, preceding and following stress, type of /s/, (6) target word length, target and following word frequency, string frequency, (7) speech rate, and social factors, which included age, gender, task, and education level (see the Appendix for a full description of the coded levels). (8) Following Garcia (2011, 2015) and Chappell & Garcia (2017), who combined the techniques of File-Muriel and Brown (2011) and Kendall (2009), rate of speech was determined by measuring the duration of the target word, preceding word, and following word and dividing the total number of syllables by time. Auditory impressions of rhotacism were confirmed via acoustic analysis of the accompanying spectrogram, with a lowered third formant and lack of high-frequency noise indicative of rhotacism. Ambiguous tokens were excluded from analysis.

Once the coding was completed, Chappell and Martinez Ibarra (2017) fitted a mixed effects binomial logistic regression model in the lme4 package (Bates et al. 2015) in R (R Core Team 2014). P-values were calculated using the lmerTest package (Kuznetsova et al. 2014). In the model construction, variant (rhotacism vs. other variant) served as the dependent variable, speaker and word were included as random effects, and a series of linguistic and social factors were tested as independent variables (see the full list in the Appendix). The best-fit mixed effects model included task, speech rate, and an interaction between gender and following segment, as shown in Section 4.

4. BEHAVIOR OF RHOTACISM IN THE DATASET. In the results of their best-fit model, the authors found an interaction between gender and following segment, which showed that men and women rhotacize differently in preconsonantal and prevocalic contexts (see Chappell & Martinez Ibarra 2017). However, two factors are particularly important to the work at hand: task and speech rate. As both independent variables significantly condition rhotacism in Elche Spanish, a more thorough exploration of their behavior will facilitate a discussion of whether or not casual and fast speech should be conflated in studies on the Spanish language. Both will be discussed after the results in Table 1.

The results of the model (Table 1) show that task was selected as a significant predictor of rhotacism in Elche Spanish; rhotacism is more likely in the informal conversation than in the reading task, which falls in line with previous research on weakened /s/ (Chappell 2015a, 2015b, Chappell & Garcia 2017, Garcia 2011,2015, among others). Speech rate was also found to significantly condition rhotacism, with rhotacism rates increasing as speech rate increases. This, too, supports previous research on /s/ weakening; Torreira and Ernestus (2011) found higher rates of /s/ voicing in faster speech, and File-Muriel and Brown (2011) find speech rate to be the most predictive factor of /s/ reduction in Caleno Spanish. (10) In sum, both casual speech and fast speech condition rhotacism in Elche Spanish. (11)

Now, if fast and casual speech do overlap and condition reductions in the same way, several predictions about the way in which they condition a reduction can be made. First of all, when casual speech is also fast speech, rhotacism should be more likely than when casual speech is slow. Similarly, when fast speech is also casual speech, rhotacism should be more likely than when fast speech is formal. To explore these predictions, I provide a two-way interaction plot in Figure 3 using Chappell and Martinez Ibarra's (2017) data. Created using the stats package in R (R Core Team 2014), Figure 3 presents possible interactions between task and speech rate by plotting the response's mean for the two factors combined. Speech rate, as measured in syllables per second, appears on the Y-axis, and the likelihood of a rhotic (0 = no, 1 = yes) appears on the X-axis. Finally, the casual sociolinguistic interview task (abbreviated as S) is shown with a solid line and the formal reading task (abbreviated as R) is presented with a dotted line.

Figure 3 demonstrates that rhotics are much more frequent in fast speech in the casual sociolinguistic interview, but the same is not true of the reading task: speech rate does not seem to condition rhotacism in the same way in the more formal task. To illustrate this point another way, Figure 4 shows the probability density of rhotics given task and speech rate in a kernel density plot. The x-axis shows speech rate, with faster speech to the right, and the y-axis shows the density of rhotics per task (reading or sociolinguistic interview). In other words, the visual clarifies where rhotic production tends to cluster within each task given speech rate.

Figure 4 helps to explain the distribution of the reduced variants based on speech rate in both formal and informal speech, and the predictions made about fast and casual speech appear to be borne out to some extent: the reduced, rhotacized /s/ occurs more in Elche Spanish when casual speech is also fast than when formal speech is fast. However, the reduction does not take place exclusively in this environment. In fact, rhotacism is found even in the slowest speaking rates and most formal speech. Additionally, the argument that coarticulation is responsible for the reduction--with higher rates of reduction in faster speech rates--is not substantiated in the formal reading task. This distribution suggests that the importance of speech rate is limited when there are greater levels of formality.

To summarize the observations made thus far, the two-way interaction plot and kernel density plot in Figures 3 and 4 indicate a stronger relationship between speech rate and task in informal speech, but the impact of speech rate is limited in formal speech. What remains to be seen is whether task or speech rate is a more useful predictor of rhotacism and whether one independent variable feeds off the other in some way. To answer this question, a conditional inference tree is presented in Figure 5 to better understand the relationship between the two. A conditional inference tree establishes whether 'an independent variable is a useful predictor of the two possible responses of the dependent variable and divides each of the predictors into subsets that are further evaluated, considering the effect of subsequent factors' (Johnson & Barnes 2013:36).

Figure 5 first selects task as the most significant predictor of rhotacism and shows that speech rate is only a useful predictor in more casual speech. In other words, speech rate's effect is minimal in formal tasks. With these findings in mind, Section 5 revisits Hasegawa's (1979, 2006) account of casual speech and fast speech in Japanese and then outlines a revised proposal to better account for rhotacism in Elche Spanish.

5. TOWARDS A REVISED APPROACH. Hasegawa (1979, 2006) has suggested that the difference between fast speech and casual speech in Japanese depends first on the applicability of speech rate. According to this approach, either fast speech affects a reduction process or it does not, and if fast speech does not affect a reduction process, a casual speech process may apply. While this approach may be the best one for reduction processes in Japanese, I contend that a simpler explanation applies in the case of Elche Spanish, potentially extendible to other dialects of Spanish and other Romance languages. The present approach places central importance on the social evaluation of a variant, or the indexical field (Eckert 2008) ascribed to a realization. Eckert (2008:453) describes the indexical field as a 'constellation of ideologically related meanings, any one of which can be activated in the situated use of the variable.' While some linguistic variants are deeply imbued with ideologically related social meanings, other variants may not be associated with any social meaning at all.

I argue that in Spanish, reduction processes are first and foremost subjected to social evaluations of contextual appropriateness, with 'casual speech' and 'fast speech' processes, as described by Hasegawa (2006), differing crucially in the prestige or stigma ascribed to the reduced form. If no stigma (or very little stigma) is associated with the lenited variant, the reduction is likely to take place in both casual and formal speech when the speech is fast, which is simply the result of greater gestural overlap with more extreme timing constraints like high vowel devoicing in Japanese. On the other hand, if a reduced variant indexes some overtly stigmatized stance, quality, or social type, that variant will likely be avoided in formal contexts and will not be as susceptible to speech rate. The strength of a variant's indexical field determines its increased use or avoidance based on context, limiting the impact of speech rate, as illustrated in Figure 6.

As depicted in Figure 6, 1 contend that social evaluations work as filters before speech rate is considered in less sociologically sensitive languages like Spanish: stigma or prestige may impede reduction rates in certain contexts, rendering a variable less susceptible to the rate of speech if speakers are more attuned to social appropriateness. If limited social judgments are associated with a variant, gestural overlap may proceed as expected, increasing in fast speech with greater timing constraints. This revised approach suggests that the first step in the determination is social, as '... even informants with a low educational level suffer from the pressure of the standard' (Hernandez-Campoy & Villena-Ponsoda 2009:206), and this pressure is heightened in more formal contexts. I also predict the effect of a variant's social evaluation and contextual appropriateness will be stronger among women, the middle class, and more educated speakers in formal settings, but all speakers will be impacted to some extent by these social judgments (see Labov 1990, 2001).

To recapitulate, I have demonstrated that /s/ rhotacism in Elche Spanish often occurs when speech is both casual and fast, drawing attention to the reason why task and speech style are so often combined in linguistic analyses: the two tend to co-occur. However, a conditional inference tree showed that task is of primary importance in determining whether or not a rhotic will occur, with speech rate only conditioning rhotacism when the speech is sufficiently informal. The importance of speech rate is effectively blocked in formal speech when social evaluations limit gestural overlap.

The findings presented here suggest that while casual speech and fast speech may overlap to some extent, they are not truly synonymous. Instead, the data from Elche Spanish suggest that social judgments act as a gatekeeper for lenition processes, with limited gestural reductions where they are deemed inappropriate. I conclude that speakers of languages like Spanish with fewer sociological sensitivities than Japanese are still highly attuned to social contexts, and that social judgments supercede coarticulatory lenition processes typical of faster speech. As a result, the conflation of casual and fast speech is not necessarily justified in linguistic analyses, and a more nuanced evaluation of each predictor's effect is needed in sociolinguistic studies to paint a more precise picture of lenition.

6. CONCLUSION. Hasegawa (1979, 2006) has shown that casual speech and fast speech processes are not the same in a highly sociologically sensitive language like Japanese, but a discussion of the difference between fast speech and casual speech has not seemed as crucial in less sociologically sensitive languages like Spanish. However, an exploration of fast speech and casual speech's effect on rhotacism in Elche Spanish has shown that fast speech and casual speech do not always overlap to the extent suggested by previous work. That is, their conflation in any sort of sociolinguistic analysis needs to be motivated by the data, not assumed as given.

Additionally, I have shown that social evaluations of appropriateness are of the utmost importance when a variant is imbued with social meaning. Strong evaluations of a variant's social and contextual appropriateness limit the importance of speech rate, but if social meaning is not linked to a reduced variant, gestural overlap may apply as expected, increasing with faster rates of speech.

The present paper contributes to the fields of phonetics and sociolinguistics by calling into question the common practice of conflating fast speech with casual speech, whose implementation is not entirely motivated based on Chappell and Martinez Ibarra's (2017) data from Elche Spanish. Instead of combining speech rate and task or exploring the impact of one but not the other, I contend that a more thorough analysis is needed to explore what the relationship is like between these two factors in any variationist study.

While the present study represents an important first step, it is not without its weaknesses. First, Elche constitutes a single dialect of Spanish, and while I have contended here that Chappell and Martinez Ibarra's (2017) results are most likely generalizable to other varieties of Spanish and even other languages, this speculation needs to be confirmed by additional studies. Would, for example, these results also be applicable for Castilian and Andalusian dialects of Spanish or Balearic Catalan where similar patterns of /s/ rhotacism occur? Would the same order of rule application apply to a different coarticulatory phenomenon that has acquired social meaning? This must be tested. Additionally, matched-guise tests are needed to more clearly identify the indexical field of /s/ rhotacism in Elche Spanish. These issues should be explored in future work to either support or refute the proposal outlined here.

In spite of these shortcomings, this work is the first to quantitatively explore the relationship between fast and casual speech in Spanish, providing a new theoretical perspective on the relationship between formality and speech rate in languages that are less sociologically sensitive than Japanese. Having called into question the tendency to analyze rate of speech and casualness as a single measure in modern sociolinguistics, sociolinguists might consider distinguishing more clearly between the two in future work, recognizing the relationship among social evaluations, formality, and speech rate.

Category                    Levels

Gender                      Female, Male
Task                        Reading, Sociolinguistic Interview
Age                         18-29, 30-49,50+
Education                   Did not finish high school, High school
                            diploma, College degree
Type of /s/                 Verbal marker, Plural marker, Lexical /s/
Target Word Class           Adjective, Adverb, Determiner, Discourse
                            marker, Noun, Preposition, Pronoun, Verb
Following Word Class        Adjective, Adverb, Conjunction, Determiner,
                            Discourse marker, Noun, Preposition,
                            Pronoun, Verb
Preceding Vowel             /a/, /e/, /i/, /o/, /u/
Following Segment           Voiceless consonant, Voiced consonant, Vowel
Preceding Syllable Stress   Unstressed, Stressed
Following Syllable Stress   Unstressed, Stressed
Preceding Word Length       1,2, 3, 4+
Speech rate                 Continuous factor, measured using syllables
                            per second
Target word frequency       Continuous, determined by the Lista total
                            de frecuencias 'Total list of frequencies'
                            from the Real Academia Espanola.
Following word frequency    Continuous, determined by the Lista total
                            de frecuencias 'Total list of frequencies'
                            from the Real Academia Espanola.
String frequency            Continuous, determined by number of word
                            1 + word 2 appearances in the present
                            corpus of spontaneous speech.


ADAMS, SCOTT G.; GARY WEISMER; and RAYMOND D. KENT. 1993. Speaking rate and speech movement velocity profiles. Journal of Speech and Hearing Research 36.41-54.

BATES, DOUGLAS; MARTIN MAECHLER; BEN BOLKER; and STEVE WALKER. 2014. Ime4: Linear mixed-effects models using Eigen and S4. R package version 1.1-12. http://CRAN.Rproject. org/package=lme4.

BECKMAN, MARY. 1996. When is a syllable not a syllable? Phonological structure and language processing: Cross-linguistic studies, ed. by Takashi Otake and Anne Cutler, 95-123. Berlin: Mouton de Gruyter.

BROWMAN, CATHERINE; and Louis GOLDSTEIN. 1990. Tiers in articulatory phonology, with some implications for casual speech. Papers in laboratory phonology I: Between the grammar and physics of speech, ed. by John Kingston and Mary E. Beckman, 341-76. Cambridge: Cambridge University Press.

BROWMAN, CATHERINE; and Louis GOLDSTEIN. 1989. Articulatory gestures as phonological units. Phonology 6.201-51.

BROWMAN, CATHERINE; and Louis GOLDSTEIN. 1986. Towards an articulatory phonology. Phonology Yearbook 3.219-52.

BROWN, ROGER; and ALBERT GILMAN. 1960. The pronouns of power and solidarity. Style in language, ed. by T. A. Sebeok, 253-76. Cambridge, MA: Technology Press of Massachusetts Institute of Technology.

CARRICABURO, NORMA. 1997. Las formulas de tratamiento en el espanol actual. Madrid: Arco/Libros.

CARVALHO, ANA MARIA. 2006. Spanish (s) aspiration as a prestige marker on the Uruguayan-Brazilian border. Spanish in Context 3.1.85-114.

CHAPPELL, WHITNEY. 2015a. Linguistic factors conditioning glottal constriction in Nicaraguan Spanish. Italian Journal of Linguistics/Rivista di Linguistica 27.2.1-42.

CHAPPELL, WHITNEY. 2015b. Formality strategies in Managua, Nicaragua: A local vs. global approach. Spanish in Context 12.2.221-54.

CHAPPELL, WHITNEY; and CHRISTINA GARCIA. 2017. Variable production and indexical social meaning: On the potential physiological origin of intervocalic /s/ voicing in Costa Rican Spanish. Studies in Hispanic and Lusophone Linguistics 10.1.1-37.

CHAPPELL, WHITNEY; and FRANCISCO MARTINEZ IBARRA. 2017. Rhotacism of /s/ in Elche Spanish: Social and linguistic factors conditioning the variant. Contemporary Advances in Theoretical and Applied Spanish Linguistic Variation, ed. by Juan Colomina-Almiflana, 43-62. Columbus, OH: The Ohio State University Press.

DOHOTARU, PUICA. 1998. /s/ final de silaba y de palabra en el habla juvenil habanera. Anuario de Linguistica Hispana XIV. 149-68.

DRESSLER, WOLFGANG; and RUTH WODAK. 1982. Sociophonological methods in the study of sociolinguistic variation in Viennese German. Language in Society 11.3.339-70.

ECKERT, PENELOPE. 2008. Variation and the indexical field. Journal of Sociolinguistics 12.3.453-76.

FILE-MURIEL, RICHARD; and EARL K. BROWN. 2011. The gradient nature of s-lenition in Caleno Spanish. Language Variation and Change 23.223-43.

GARCIA, CHRISTINA. 2011. Intervocalic /s/ voicing in the Spanish of Loja, Ecuador. Columbus, OH: The Ohio State University MA thesis.

GARCIA, CHRISTINA. 2015. Gradience and variability of intervocalic /s/ voicing in Highland Ecuadorian Spanish. Columbus, OH: The Ohio State University PhD dissertation.

GIMENO-MENENDEZ, FRANCISCO; and JOSE R. GOMEZ MOLINA. 2007. Spanish and Catalan in the Community of Valencia. International Journal of the Sociology of Language 184.95-107.

HASEGAWA, NOBUKO. 2006. On casual speech: How it differs from fast speech. Perspectives on teaching connected speech to second language speakers, ed. by Kimi Kondo-Brown, 169-87. Honolulu, HI: Second Language Teaching & Curriculum Center, University of Hawai'i at Manoa.

HASEGAWA, NOBUKO. 1979. Casual speech vs. fast speech. Proceedings of the Chicago Linguistics Society 15.126-37.

HERNANDEZ-CAMPOY, JUAN MANUEL; and JUAN ANDRES VILLENA-PONSODA. 2009. Standardness and non-standardness in Spain: Dialect attrition and revitalisation of regional dialects of Spanish. International Journal of the Sociology of Language .181-214.

INSTITUTO NACIONAL DE ESTADISTICA. 2013. INEbase, Censos de poblacion y viviendas. Online:

JACEWICZ, EWA; and ROBERT A. Fox. 2010. Between-speaker and within-speaker variation in speech tempo of American English. Journal of the Acoustical Society of America 128.2.839-50.

JOHNSON, KEITH. 2004. Massive reduction in conversational American English. Spontaneous speech: Data and analysis. Proceedings of the 1st Session of the 10th International Symposium, ed. by Kiyoko Yoneyama and Kikuo Maekawa, 29-54. Tokyo: The National International Institute for Japanese Language.

JOHNSON, MARY; and SONIA BARNES. 2013. Haya vs. haiga: An analysis of the variation observed in Mexican Spanish using a mixed effects model. Selected Proceedings of the 6th Workshop on Spanish Sociolinguistics, ed. by Ana M. Carvalhoand Sara Beaudrie, 32-40. Somerville, MA: Cascadilla Proceedings Project.

JUN, SUN-AH; and MARY E. BECKMAN. 1993. A gestural-overlap analysis of vowel devoicing in Japanese and Korean. Paper presented at the annual meeting of the Linguistic Society of America, Los Angeles.

KAPOVTC, MARKO. 2015. Rhotacism in Spanish and its decline in Ciudad Real (Spain): A case study of a change in progress nearing completion. Sociolinguistic Studies 9.4.399-419.

KENDALL, TYLER. 2009. Speech rate, pause, and linguistic variation: An examination through the Sociolinguistic Archive and Analysis Project. Durham, NC: Duke University dissertation.

KIM, JONG-BOK; and PETER SELLS. 2007. Korean honorification: a kind of expressive meaning. Journal of East Asian Linguistics 16.4.303-36.

KUEHN, DAVID P., and KENNETH L. MOLL. 1976. A cineflourographic investigation of CV and VC articulatory velocities. Journal of Phonetics 3.303-20.

KUZNETSOVA, ALEXANDRA; PER B. BROCKHOFF; and RUNE H. B. CHRISTENSEN. (2014). lmerTest: Tests for random and fixed effects for linear mixed effect models (lmer objects of Ime4 package). R package version 2.0-6. Online:

LABOV, WILLIAM. 1964. Phonological correlates of social stratification. American Anthropologist

LABOV, WILLIAM. 1972. Sociolinguistic patterns. Philadelphia: University of Pennsylvania Press.

LABOV, WILLIAM. 1990. The intersection of sex and social class in the course of linguistics change. Language Variation and Change 2.205-54.

LABOV, WILLIAM. 2001. Principles of Linguistic Change: Social Factors. Oxford: Blackwell.

LINDBLOM, BJORN. 1963. Spectrographic study of vowel reduction. Journal of the Acoustic Society of America 35.1773-81.

LIPSKI, JOHN. 1985. /s/ in Central American Spanish. Hispania 68.143-49.

LUBKER, JAMES; and THOMAS GAY. 1982. Anticipatory labial coarticulation: Experimental, biological, and linguistic variables. Journal of the Acoustical Society of America 71.437-48.

MARTINEZ IBARRA, FRANCISCO. 2013. Why would I speak Valencian? Understanding language rejection in the southern territories of the Autonomous Community of Valencia. International Journal of the Linguistic Association of the Southwest 32.1.67-99.

MATTHIES, MELANIE; PASCAL PERRIER; JOSEPH S. PERKELL; and MAJID ZANDIPOUR. 2001. Variation in anticipatory coarticulation with changes in clarity and rate. Journal of Speech, Language, and Hearing Research 44.340-53.

MILROY, JAMES. 2001. Language ideologies and the consequences of standardization. Journal of Sociolinguistic s 5.530-55.

MILROY, LESLEY. 1980. Language and social networks. Oxford: Blackwell.

MORGAN, TERRELL A.; and SCOTT SCHWENTER. 2016. Vosotros, Ustedes, and the Myth of the Symmetrical Castilian Pronoun System. In Inquiries in Hispanic Linguistics: From theory to empirical evidence, ed. by Alejandro Cuza, Lori Czerwionka and Daniel Olson, 263-80. Philadelphia: John Benjamins.

NOLAN, FRANCIS. 1991. The descriptive role of segments: Evidence from assimilation. Papers in laboratory phonology II: Gesture, segment, prosody, ed. by Gerard J. Docherty and D. Robert Ladd, 261-80. Cambridge: Cambridge University Press.

PAINTER, ROBERT KENNETH. 2011. Acoustic and perceptual explanations for rhotacism in Latin and Germanic. Buffalo, NY: State University of New York dissertation.

PAREDES GARCIA, FLORENTINO. 2001. El habla de la Jara. Los sonidos. Alcala de Henares: Universidad de Alcala.

PARK, MAE-RAN. 1990. Conflict avoidance in social interaction: A sociolinguistic comparison of the Korean and Japanese honorific systems. Japanese/Korean linguistics, ed. by Hajime Hoji, 111-27. Stanford, CA: Center for the Study of Language and Information.

R CORE TEAM. 2014. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Online: URL

REISCHAUER, EDWIN O. 1995. The Japanese today: Change and continuity. Cambridge: Harvard University Press.

ROBERTS, PHILIP J. 2012. Latin rhotacism: A case study in the life cycle of phonological processes. Transactions of the Philological Society 110.1.80-93.

ROMERO, JOAQUIN; and SIDNEY MARTIN. 2003. Articulatory weakening as basis of historical rhotacism. Proceedings of 15th International Congress of Phonetics Sciences, ed. by Maria-Josep Sole, Daniel Recasens, and Joaquin Romero, 2825-28. Barcelona: Causal Productions.

RUIZ-SANCHEZ, CARMEN. 2004. El comportamiento de la /s/ implosiva en el habla de Caracas. Boletin de linguistica 21.48-65.

SHAIMAN, SUSAN; SCOTT G. ADAMS; and MIKAEL O. Z. KIMELMAN. 1997. Velocity profiles of lip protrusion across changes in speaking rate. Journal of Speech, Language, and Hearing Research 40.144-58.

SINNOTT, SARAH T. 2010. Address forms in Castilian Spanish: Convention and implica-ture. Columbus, OH: The Ohio State University dissertation.

SMITH, NEILSON V. 1973. The acquisition of phonology: A case study. Cambridge: Cambridge University Press.

SOLE, MARIA JOSEP. 1992. Experimental phonology: The case of rhotacism. Phonologica 1988: Proceedings of the 6th International Phonology Meeting. Cambridge: Cambridge University Press.

SONG, JUYOUNG. 2010. Bilingual creativity and self-negotiation: Korean American children's language socialization into Korean address terms. Beyond Yellow English: Towards linguistic anthropology of Asian Pacific America, ed. by Angela Reyes and Adrienne Lo, 213-31. Oxford University Press.

STEWART, MIRANDA. 1999. The Spanish language today. London: Routledge.

TAUROZA, STEVE; and DESMOND ALLISON. 1990. Speech rates in British English. Applied Linguistics 11.1.90-109.

TERRELL, TRACY. 1979. Final /s/ in Cuban Spanish. Hispania 62.4.599-612.

TORREIRA, FRANCISCO; and MIRJAM ERNESTUS. 2011. Voicing of intervocalic /s/ in conversational Spanish. Paper presented at Phonetics and Phonology in Iberia, Tarragona, Spain.

WHEELER, MAX; ALAN YATES; and NICOLAU DOLS. 1999. Catalan: A comprehensive grammar. New York: Routledge.

ZWICKY, ARNOLD. 1972. On casual speech. Papers from the Eighth Regional Meeting of the Chicago Linguistics Society, ed. by Paul Peranteau, Judith Levi, and Gloria Phares, 607-15. Chicago: Chicago Linguistics Society.

Department of Modern Languages and Literatures

The University of Texas at San Antonio

San Antonio, Texas 78249



The University of Texas at San Antonio

(*) I am grateful to Paco Martinez Ibarra, my friend and colleague, for granting me permission to use his corpus of Elche Spanish. I also thank four anonymous reviewers for their helpful feedback on an earlier version of this paper. Any remaining mistakes are mine alone.

(1) Hasegawa (2006) clarifies that fast speech processes 'are observed in even slow or careful speech; however, they apply much more freely and across word-boundaries in fast speech' (170), while casual speech processes' application ' not affected by the rate of speech. They apply when the context of speech is casual or informal enough' (173). None of the processes described by Hasegawa appear to be categorical in their application but rather much more frequent in fast speech or in casual contexts, respectively.

(2) no serves as either a sentence-final nominalizer or a possessive/genitive particle.

(3) ni is an adverbial/connective.

(4) Labov ( 1990, 2001 ) argues that men tend to use nonstandard variants more than women in cases of stable variation and changes from above. Therefore, it is not entirely unexpected that men would engage more frequently in casual speech reductions than women.

(5) It is not clear if /s/ rhotacism is stigmatized in other regions of Spain, e.g. Ciudad Real (see Kapovic 2015) or La Jara (see Paredes Garcia 2001:141).

(6) This category included /s/ as a verbal marker, e.g., ladras 'you bark,' /s/ as a plural marker, e.g., perros 'dogs,' and lexical /s/, e.g., mas 'more'

(7) Both target and following word frequency were measured using the Lista total de frecuencias 'Total list of frequencies' available through the Real Academia Espanola 'Royal Spanish Academy." String frequency was determined using the present spontaneous speech corpus given the number of word 1 + word 2 appearances in the recordings.

(8) Following Chappell (2015a, 2015b), the reading task was conducted after the sociolinguistic interview to maximally decrease perceptions of formality in the first task.

(9) See Chappell and Martinez Ibarra (2017) for a more detailed discussion of the model.

(10) While /s/ aspiration and elision are not as clearly the result of coarticulation, File-Muriel and Brown's (2011) work is mentioned here as it involves another /s/-related reduction process.

(11) Given the association between fast speech and casual speech, separate models were also fitted to rhotacism to avoid the inclusion of both independent variables in a single model. The same results were found for these more minimalistic models: rhotacism is more likely in faster speech, on the one hand, and in more casual tasks, on the other (p < 0.01).
TABLE 1. Significant factors conditioning rhotacism in Elche (from
Chappell & Martinez Ibarra 2017: 57).9

                     Estimate   Std.                   z-value  p-value

(Intercept)          -6.53673   1.09763                -5.955    <01
Gender: Following
Segment Interaction
(reference level
is female,
voiced consonant)
Male: Following       2.19961   0.77488                 2.839    <01
Task (reference
level is reading)
Spontaneous           1.86459   0.53896                 3.460   <.01
Speech Rate
Faster speech         0.19160   0.07344                 2.609    <01
AIC 518.7            BIC 562.8  Log likelihood -250.3

[Please note: Some non-Latin characters were omitted from this article]
COPYRIGHT 2014 Linguistic Association of the Southwest
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2014 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Chappell, Whitney
Publication:International journal of the Linguistic Association of the Southwest
Geographic Code:4EUSP
Date:Dec 1, 2014

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters