Single Case ReportSaying thirteen instead of forty-two but saying lale instead of tale: is number production special?
Introduction
The study of word production skills in aphasics has been a primary source of information in the development of theoretical models of word production (e.g., Hillis, 2001). Recently, interest has arisen in a paradoxical phenomenon that Dotan and Friedmann (2015) have called Stimulus Type Effect on Phonological and Semantic errors (STEPS). This phenomenon, to the best of our knowledge, was first described by Geschwind (1965) and alludes to the existence of more phonemic than semantic errors in tasks that involve the production of non-number words (e.g., lale for tale) and pseudowords (e.g., petal for pepal), but more semantic than phonemic errors when producing number words (e.g., thirteen for forty-two). This pattern is intriguing because, according to classical models of word production (e.g., Butterworth, 1992; Ellis, 1980; Levelt, 1989, Levelt et al., 1999), phonemic paraphasias, in that they emerge peripherically, should affect the production of any kind of word, including numbers (e.g., Shallice, Rumiati, & Zadini, 2000). At the same time, it is difficult to explain, in the absence of impairment in the semantic representations of quantity, why semantic errors are restricted to the category of numbers. Importantly, from a theoretical perspective this phenomenon has been cited recently in support of the claim that number words are special and are produced differently from other word categories (e.g., Bencini et al., 2011, Dotan and Friedmann, 2015, Dotan and Friedmann, 2019).
The first experimental description of STEPS can be found in a paper by Cohen, Verstichel, and Dehaene (1997). They described the case of a 76-year-old retired college teacher diagnosed with Wernicke's aphasia who showed very clear STEPS in naming and reading words (because he suffered from cortical deafness his repetition was not assessed). He produced severe neologistic jargon during naming and word reading (e.g., alugre for araignée), whereas in naming Arabic numerals or reading number words he produced almost exclusively semantic (e.g., 460 for 250) and syntactic (e.g., 704 for 74) errors. That is, in the non-number words, errors affected the selection of phonemes, whereas in number words, errors affected whole words, thus giving rise to the substitution of a number word for another number word. Cohen et al. (1997) also found a prevalence of substitution errors in letter naming (e.g., saying /ϵm/ for G). To explain this pattern, they argued that numbers, letters and phonemes have in common the fact that they are basic units or building blocks of language, used to generate more complex sequences (i.e., multi-digits, acronyms and words, respectively) (Cohen et al., 1997). Although they recognize that such a theory is highly speculative, it does provide a clear explanation for: a) the occurrence of phonemic errors in non-number words, which would thus originate in an impairment in the selection of phonemes that are a word's building blocks; and b) the presence of semantic errors in numbers, which would thus be caused by an impairment in the selection of a number's building blocks (Cohen et al., 1997).
STEPS has also been studied by Messina, Denes, and Basso (2009) and Ochtrup et al. (2013), respectively, in groups of 57 and 15 patients with different types of aphasia, during naming and repetition tasks. The existence of the phenomenon has also been noted and indeed described in other studies, although not as the main focus of interest (see Bachoud-Lévi and Dupoux, 2003, Bencini et al., 2011, Delazer and Bartha, 2001, Girelli and Delazer, 1999, Marangolo et al., 2004, Marangolo et al., 2005; Rodriguez & Laganaro, 2008; Shallice et al., 2000).
However to our knowledge, the most exhaustive study on STEPS was carried out recently by Dotan and Friedmann (2015). They provide a detailed description of the effect in 6 aphasic patients who all had an impairment in the phonological output buffer (POB), that is, the cognitive deficit usually underlying conduction aphasia of the “reproduction” variety (e.g., Shallice et al., 2000, Shallice and Warrington, 1977). Dotan and Friedmann (2015) found that, in all 6 patients, errors in non-number words were predominantly phonemic, whereas in number words they were predominantly syntactic (e.g., saying 506 when presented with 560) and semantic (e.g., saying 48 when presented with 56). This pattern did not depend on the nature of the task, since it was observed in repetition, naming, and reading tasks. On the rationale that building blocks are preassembled phonological units larger than a single phoneme, and that they are used “in productive phonological process that assembles them into a more complex phonological structure” (Dotan & Friedmann, 2015, p. 333), they looked at whether stimuli other than number and letter names, specifically function words and morphological affixes, might also be viewed as building blocks. Using different tasks, they found the whole substitution of function words and morphological affixes, thus confirming their hypothesis and also pointing to the relevance of the role in which a word appears: “When number words and function words appear in the relevant role (number words with numeric meaning and function words with syntactic role), they are produced with semantic rather than phonological errors. Conversely, when the number and function words were deprived of their role by changing the task and context, they were produced with many phonological errors, and without semantic errors, just like content words” (Dotan & Friedmann, 2015, p. 338).
Dotan and Friedmann interpret their findings as support for the existence in the POB of several mini-stores that contain the pre-assembled phonological forms of single phonemes, single number words, function words, morphological affixes and letter names (see also Fig. 1). These units act as the building blocks of more complex productions such as words, multi-digit numbers, sentences, morphological complex words and acronyms, respectively (Dotan & Friedmann, 2015). From this perspective, impairments in the POB would lead to patients substituting the target item with another within the same mini-store, thus producing: a phonemic error in the case of words; an ‘apparently’ semantic error in the case of numbers; a morphological error in the case of morphological complex words; the substitution of a function word in sentences; and substitution errors in naming letters. With this interpretation Dotan and Friedmann refine the hypothesis originally proposed by Cohen et al. (1997).
Within the framework of serial production models, such as that used by Dotan and Friedmann (2015) and Cohen et al. (1997), it is difficult to understand how semantic and lexical distinctions like number, letters, function words, etc., are projected onto the POB, a system that stores phonological information and thus should be organized according to these properties. It is also difficult to imagine how the POB “knows” when a phonological sequence, such as the one corresponding to a function word, will be used in the frame of a sentence or in a word list, thus activating a preassembled “block” or assembling a sequence of phonemes from scratch. Moreover, the literature on picture and Arabic number naming in children (errors are unusual in healthy adults) indicates that the most common types of error in naming words are semantic, whereas in naming Arabic numbers they are syntactic, these usually disappearing early on, followed at a considerable distance by the disappearance of lexical errors (note that Dotan & Friedmann considered both as semantic errors). Interestingly, evidence of phonemic errors is scant in both cases, indeed, as scant as in the case of semantic errors in numbers accompanied by phonemic errors in non-number words (Barrouillet et al., 2004, McGregor et al., 2002;Moura et al., 2013). The pattern found in development thus suggests the existence of similar mechanisms for both types of stimuli, making the existence of dedicated mechanisms for non-number and number words implausible; more importantly, it also suggests that the system where the semantic errors in numbers occur is, as in the case of non-number words, the semantic level, not the POB.
Finally, although this has not been explored in depth, there is some evidence suggesting that STEPS may appear in patients without impairment in the POB. As noted above, Messina et al. (2009) describe 57 patients, mainly with Broca's and Wernicke's aphasias, and find that the effect was found when considering these as a group. However, impairments in the POB are not usual in the latter type of patients (see Robson, Sage, & Lambon Ralph, 2012). Also, and more importantly, the present study is based on the clinical observation of a patient who, despite the preservation of the POB, seems to show STEPS.
Thus, it seems that despite the general support that Dotan and Friedmann's hypothesis enjoys, the basic assumption of their model, this is, that the STEPS effect emerges as a consequence of the impairment of the POB ‒ the repository of the building blocks: phonemes for content words, and number words for multi-digits ‒ has not previously been tested. This, then, is the main objective of the present study.
With the above brief survey in mind, the aim of this study is to test whether STEPS has its origin in an impairment of the POB. To explore the issue, we evaluated two conduction aphasia patients. Patient DNR suffers mainly from a conduction aphasia of the reproduction variety, and this corresponds to an impairment of the POB, thus causing difficulties in repetition, naming, and reading (e.g., Caplan et al., 1986, Caramazza et al., 1986, García-Orza and León-Carrión, 2005, Shallice et al., 2000, Shallice and Warrington, 1977, Tomasino et al., 2015. The Building Blocks hypothesis would predict STEPS in this patient. Patient ML suffers what has been termed conduction aphasia of the repetition variety, and this corresponds to an impairment in the phonological input buffer (PIB), thus leading to problems mainly with repetition but not with naming or reading (e.g., Martin et al., 1999, Shallice and Warrington, 1977, Sidiropoulos et al., 2008). According to the Building Blocks hypothesis, no STEPS should be observed in this patient, in that her POB is preserved.
We therefore explored whether STEPS is limited to the POB patient, as predicted by the Building Block hypothesis, or whether it also appears in the PIB patient. To this aim we evaluated and compared, in both patients, the distribution of semantic and phonemic errors in the repetition, naming, and reading of number and non-number words. Although considerable debate exists as to whether Arabic numbers are produced as words or as pictures (see Brysbaert, 2005 for a review), here we compared performance in Arabic number reading with that in picture naming, and performance in reading numbers presented in orthographic code with that in word reading (see more on this in the section 4).
Section snippets
Participants
Two female patients took part in the study, which was carried out between January and December 2018. During the testing period, their general neuropsychological condition remained unchanged. The research was conducted in accordance with the Code of Ethics of the World Medical Association (Declaration of Helsinki) for experiments involving humans. The study was approved by the Local Ethic Committee of Universidad de Málaga. Informed consent was obtained from both participants.
The study: STEPS evaluation
In order to evaluate STEPS in DNR (impairment of the POB) and ML (impairment of the PIB) we developed a set of specific tasks. Summarizing, the patients were required to repeat, read, and name 60 non-number words and 60 numbers. A complete description of the tasks used is provided below.
Discussion
Investigating patients' neuropsychological symptoms has proven to be a fruitful approach for both clinicians and cognitive researchers (Caramazza, 1986, Coltheart et al., 2001, Schwartz and Dell, 2010). The present study sought to evaluate a theoretical explanation for a paradoxical behavior found in many aphasic patients: the production of more phonemic paraphasias when asked to produce content words, but the production of considerably more semantic errors when asked to produce numbers, the
Conclusion
The data presented here cast doubt on the Building Blocks hypothesis as an explanation of the notion that STEPS is exclusively a consequence of an impairment in the POB. A modification of the original proposal, in which the PIB is considered to be analogous with the POB, and would thus include mini-stores for pre-assembled phonological sequences, led to additional problems, such as the need to explain how semantic or contextual factors influence the selection of the mini-stores that have to
Authors' contribution
JGO: Conceptualization, Investigation, Formal Analysis, Writing-Original Draft, Writing-Review & Editing; Funding Acquisition; IGC: Data Curation, Formal Analysis, Investigation, Writing-Original Draft, Visualization; MG: Data Curation, Investigation, Writing-Original Draft.
Open practices
The study in this article earned an Open Data badge for transparent practices. Materials and data for the study are available at https://osf.io/y6pxg.
In this repository the pictures employed in the naming task are not included as the images are protected by copyright (Duñabeitia et al., 2018), however, the reference number to identify each picture in the data base is provided. No part of the study procedures or analysis plans was preregistered prior to the research being conducted. We report
Declaration of Competing Interest
None declared.
Acknowledgements
We would like to thank DNR, ML and their relatives for their collaboration in this study. We also thank Roberto Cubelli, Giorgio Arcara, an anonymous reviewer, José Miguel Rodríguez-Santos, Marcelo Berthier and particularly Diana López-Barroso for their inspiring comments on this research. The study was supported by grants from the Universidad de Malaga-Andalucía Tech which were awarded to MG and IGC, respectively. This research was also partially funded by a grant from the Ministerio de
References (67)
- et al.
When two and too don't go together: A selective phonological deficit sparing number words
Cortex
(2011) Disorders of phonological encoding
Cognition
(1992)On drawing inferences about the structure of normal cognitive systems from the analysis of patterns of impaired performance
Brain and Cognition
(1986)- et al.
Effects of semantic context in the naming of pictures and words
Cognition
(2001) - et al.
Steps towards understanding the phonological output buffer and its role in the production of numbers, morphemes, and function words
Cortex
(2015) - et al.
Separate mechanisms for number reading and word reading: Evidence from selective impairments
Cortex
(2019) Errors in speech and short-term memory: The effects of phonemic similarity and syllable position
Journal of Verbal Learning and Verbal Behavior
(1980)- et al.
Lexical effects in verbal STM: Evidences from a phonological output buffer patient
Brain and Language
(2005) - et al.
Phonologically related lexical repetition disorder: A case study
Brain and Language
(2001) - et al.
Lexical and post-lexical phonological representations in spoken production
Cognition
(2007)